| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1 | <html><body> |
| 2 | <style> |
| 3 | |
| 4 | body, h1, h2, h3, div, span, p, pre, a { |
| 5 | margin: 0; |
| 6 | padding: 0; |
| 7 | border: 0; |
| 8 | font-weight: inherit; |
| 9 | font-style: inherit; |
| 10 | font-size: 100%; |
| 11 | font-family: inherit; |
| 12 | vertical-align: baseline; |
| 13 | } |
| 14 | |
| 15 | body { |
| 16 | font-size: 13px; |
| 17 | padding: 1em; |
| 18 | } |
| 19 | |
| 20 | h1 { |
| 21 | font-size: 26px; |
| 22 | margin-bottom: 1em; |
| 23 | } |
| 24 | |
| 25 | h2 { |
| 26 | font-size: 24px; |
| 27 | margin-bottom: 1em; |
| 28 | } |
| 29 | |
| 30 | h3 { |
| 31 | font-size: 20px; |
| 32 | margin-bottom: 1em; |
| 33 | margin-top: 1em; |
| 34 | } |
| 35 | |
| 36 | pre, code { |
| 37 | line-height: 1.5; |
| 38 | font-family: Monaco, 'DejaVu Sans Mono', 'Bitstream Vera Sans Mono', 'Lucida Console', monospace; |
| 39 | } |
| 40 | |
| 41 | pre { |
| 42 | margin-top: 0.5em; |
| 43 | } |
| 44 | |
| 45 | h1, h2, h3, p { |
| 46 | font-family: Arial, sans serif; |
| 47 | } |
| 48 | |
| 49 | h1, h2, h3 { |
| 50 | border-bottom: solid #CCC 1px; |
| 51 | } |
| 52 | |
| 53 | .toc_element { |
| 54 | margin-top: 0.5em; |
| 55 | } |
| 56 | |
| 57 | .firstline { |
| 58 | margin-left: 2 em; |
| 59 | } |
| 60 | |
| 61 | .method { |
| 62 | margin-top: 1em; |
| 63 | border: solid 1px #CCC; |
| 64 | padding: 1em; |
| 65 | background: #EEE; |
| 66 | } |
| 67 | |
| 68 | .details { |
| 69 | font-weight: bold; |
| 70 | font-size: 14px; |
| 71 | } |
| 72 | |
| 73 | </style> |
| 74 | |
| 75 | <h1><a href="bigquery_v2.html">BigQuery API</a> . <a href="bigquery_v2.models.html">models</a></h1> |
| 76 | <h2>Instance Methods</h2> |
| 77 | <p class="toc_element"> |
| 78 | <code><a href="#delete">delete(projectId, datasetId, modelId)</a></code></p> |
| 79 | <p class="firstline">Deletes the model specified by modelId from the dataset.</p> |
| 80 | <p class="toc_element"> |
| 81 | <code><a href="#get">get(projectId, datasetId, modelId)</a></code></p> |
| 82 | <p class="firstline">Gets the specified model resource by model ID.</p> |
| 83 | <p class="toc_element"> |
| 84 | <code><a href="#list">list(projectId, datasetId, pageToken=None, maxResults=None)</a></code></p> |
| 85 | <p class="firstline">Lists all models in the specified dataset. Requires the READER dataset</p> |
| 86 | <p class="toc_element"> |
| 87 | <code><a href="#list_next">list_next(previous_request, previous_response)</a></code></p> |
| 88 | <p class="firstline">Retrieves the next page of results.</p> |
| 89 | <p class="toc_element"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 90 | <code><a href="#patch">patch(projectId, datasetId, modelId, body=None)</a></code></p> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 91 | <p class="firstline">Patch specific fields in the specified model.</p> |
| 92 | <h3>Method Details</h3> |
| 93 | <div class="method"> |
| 94 | <code class="details" id="delete">delete(projectId, datasetId, modelId)</code> |
| 95 | <pre>Deletes the model specified by modelId from the dataset. |
| 96 | |
| 97 | Args: |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 98 | projectId: string, Required. Project ID of the model to delete. (required) |
| 99 | datasetId: string, Required. Dataset ID of the model to delete. (required) |
| 100 | modelId: string, Required. Model ID of the model to delete. (required) |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 101 | </pre> |
| 102 | </div> |
| 103 | |
| 104 | <div class="method"> |
| 105 | <code class="details" id="get">get(projectId, datasetId, modelId)</code> |
| 106 | <pre>Gets the specified model resource by model ID. |
| 107 | |
| 108 | Args: |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 109 | projectId: string, Required. Project ID of the requested model. (required) |
| 110 | datasetId: string, Required. Dataset ID of the requested model. (required) |
| 111 | modelId: string, Required. Model ID of the requested model. (required) |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 112 | |
| 113 | Returns: |
| 114 | An object of the form: |
| 115 | |
| 116 | { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 117 | "location": "A String", # Output only. The geographic location where the model resides. This value |
| 118 | # is inherited from the dataset. |
| 119 | "friendlyName": "A String", # Optional. A descriptive name for this model. |
| 120 | "lastModifiedTime": "A String", # Output only. The time when this model was last modified, in millisecs since the epoch. |
| 121 | "labels": { # The labels associated with this model. You can use these to organize |
| 122 | # and group your models. Label keys and values can be no longer |
| 123 | # than 63 characters, can only contain lowercase letters, numeric |
| 124 | # characters, underscores and dashes. International characters are allowed. |
| 125 | # Label values are optional. Label keys must start with a letter and each |
| 126 | # label in the list must have a different key. |
| 127 | "a_key": "A String", |
| 128 | }, |
| 129 | "labelColumns": [ # Output only. Label columns that were used to train this model. |
| 130 | # The output of the model will have a "predicted_" prefix to these columns. |
| 131 | { # A field or a column. |
| 132 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 133 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 134 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 135 | # in this case the output parameter does not have this "type" field). |
| 136 | # Examples: |
| 137 | # INT64: {type_kind="INT64"} |
| 138 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 139 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 140 | # {type_kind="STRUCT", |
| 141 | # struct_type={fields=[ |
| 142 | # {name="x", type={type_kind="STRING"}}, |
| 143 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 144 | # ]}} |
| 145 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 146 | "fields": [ |
| 147 | # Object with schema name: StandardSqlField |
| 148 | ], |
| 149 | }, |
| 150 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 151 | "typeKind": "A String", # Required. The top level type of this field. |
| 152 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 153 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 154 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 155 | ], |
| 156 | "modelType": "A String", # Output only. Type of the model resource. |
| 157 | "featureColumns": [ # Output only. Input feature columns that were used to train this model. |
| 158 | { # A field or a column. |
| 159 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 160 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 161 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 162 | # in this case the output parameter does not have this "type" field). |
| 163 | # Examples: |
| 164 | # INT64: {type_kind="INT64"} |
| 165 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 166 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 167 | # {type_kind="STRUCT", |
| 168 | # struct_type={fields=[ |
| 169 | # {name="x", type={type_kind="STRING"}}, |
| 170 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 171 | # ]}} |
| 172 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 173 | "fields": [ |
| 174 | # Object with schema name: StandardSqlField |
| 175 | ], |
| 176 | }, |
| 177 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 178 | "typeKind": "A String", # Required. The top level type of this field. |
| 179 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 180 | }, |
| 181 | }, |
| 182 | ], |
| 183 | "expirationTime": "A String", # Optional. The time when this model expires, in milliseconds since the epoch. |
| 184 | # If not present, the model will persist indefinitely. Expired models |
| 185 | # will be deleted and their storage reclaimed. The defaultTableExpirationMs |
| 186 | # property of the encapsulating dataset can be used to set a default |
| 187 | # expirationTime on newly created models. |
| 188 | "trainingRuns": [ # Output only. Information for all training runs in increasing order of start_time. |
| 189 | { # Information about a single training query run for the model. |
| 190 | "startTime": "A String", # The start time of this training run. |
| 191 | "results": [ # Output of each iteration run, results.size() <= max_iterations. |
| 192 | { # Information about a single iteration of the training run. |
| 193 | "trainingLoss": 3.14, # Loss computed on the training data at the end of iteration. |
| 194 | "evalLoss": 3.14, # Loss computed on the eval data at the end of iteration. |
| 195 | "index": 42, # Index of the iteration, 0 based. |
| 196 | "learnRate": 3.14, # Learn rate used for this iteration. |
| 197 | "durationMs": "A String", # Time taken to run the iteration in milliseconds. |
| 198 | "arimaResult": { # (Auto-)arima fitting result. Wrap everything in ArimaResult for easier |
| 199 | # refactoring if we want to use model-specific iteration results. |
| 200 | "arimaModelInfo": [ # This message is repeated because there are multiple arima models |
| 201 | # fitted in auto-arima. For non-auto-arima model, its size is one. |
| 202 | { # Arima model information. |
| 203 | "arimaFittingMetrics": { # ARIMA model fitting metrics. # Arima fitting metrics. |
| 204 | "aic": 3.14, # AIC. |
| 205 | "logLikelihood": 3.14, # Log-likelihood. |
| 206 | "variance": 3.14, # Variance. |
| 207 | }, |
| 208 | "timeSeriesId": "A String", # The id to indicate different time series. |
| 209 | "arimaCoefficients": { # Arima coefficients. # Arima coefficients. |
| 210 | "movingAverageCoefficients": [ # Moving-average coefficients, an array of double. |
| 211 | 3.14, |
| 212 | ], |
| 213 | "autoRegressiveCoefficients": [ # Auto-regressive coefficients, an array of double. |
| 214 | 3.14, |
| 215 | ], |
| 216 | "interceptCoefficient": 3.14, # Intercept coefficient, just a double not an array. |
| 217 | }, |
| 218 | "hasDrift": True or False, # Whether Arima model fitted with drift or not. It is always false |
| 219 | # when d is not 1. |
| 220 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported |
| 221 | # for one time series. |
| 222 | "A String", |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 223 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 224 | "nonSeasonalOrder": { # Arima order, can be used for both non-seasonal and seasonal parts. # Non-seasonal order. |
| 225 | "q": "A String", # Order of the moving-average part. |
| 226 | "d": "A String", # Order of the differencing part. |
| 227 | "p": "A String", # Order of the autoregressive part. |
| 228 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 229 | }, |
| 230 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 231 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported for |
| 232 | # one time series. |
| 233 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 234 | ], |
| 235 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 236 | "clusterInfos": [ # Information about top clusters for clustering models. |
| 237 | { # Information about a single cluster for clustering model. |
| 238 | "clusterSize": "A String", # Cluster size, the total number of points assigned to the cluster. |
| 239 | "centroidId": "A String", # Centroid id. |
| 240 | "clusterRadius": 3.14, # Cluster radius, the average distance from centroid |
| 241 | # to each point assigned to the cluster. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 242 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 243 | ], |
| 244 | }, |
| 245 | ], |
| 246 | "evaluationMetrics": { # Evaluation metrics of a model. These are either computed on all training # The evaluation metrics over training/eval data that were computed at the |
| 247 | # end of training. |
| 248 | # data or just the eval data based on whether eval data was used during |
| 249 | # training. These are not present for imported models. |
| 250 | "binaryClassificationMetrics": { # Evaluation metrics for binary classification/classifier models. # Populated for binary classification/classifier models. |
| 251 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 252 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 253 | # macro-averaged, the metrics are calculated for each label and then an |
| 254 | # unweighted average is taken of those values. When micro-averaged, the |
| 255 | # metric is calculated globally by counting the total number of correctly |
| 256 | # predicted rows. |
| 257 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 258 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 259 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 260 | # classification models this is the positive class threshold. |
| 261 | # For multi-class classfication models this is the confidence |
| 262 | # threshold. |
| 263 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 264 | # metric. |
| 265 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 266 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 267 | # this is a macro-averaged metric. |
| 268 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 269 | # positive actual labels. For multiclass this is a macro-averaged |
| 270 | # metric treating each class as a binary classifier. |
| 271 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 272 | # multiclass this is a micro-averaged metric. |
| 273 | }, |
| 274 | "negativeLabel": "A String", # Label representing the negative class. |
| 275 | "positiveLabel": "A String", # Label representing the positive class. |
| 276 | "binaryConfusionMatrixList": [ # Binary confusion matrix at multiple thresholds. |
| 277 | { # Confusion matrix for binary classification models. |
| 278 | "falseNegatives": "A String", # Number of false samples predicted as false. |
| 279 | "falsePositives": "A String", # Number of false samples predicted as true. |
| 280 | "trueNegatives": "A String", # Number of true samples predicted as false. |
| 281 | "f1Score": 3.14, # The equally weighted average of recall and precision. |
| 282 | "precision": 3.14, # The fraction of actual positive predictions that had positive actual |
| 283 | # labels. |
| 284 | "positiveClassThreshold": 3.14, # Threshold value used when computing each of the following metric. |
| 285 | "accuracy": 3.14, # The fraction of predictions given the correct label. |
| 286 | "truePositives": "A String", # Number of true samples predicted as true. |
| 287 | "recall": 3.14, # The fraction of actual positive labels that were given a positive |
| 288 | # prediction. |
| 289 | }, |
| 290 | ], |
| 291 | }, |
| 292 | "regressionMetrics": { # Evaluation metrics for regression and explicit feedback type matrix # Populated for regression models and explicit feedback type matrix |
| 293 | # factorization models. |
| 294 | # factorization models. |
| 295 | "meanSquaredError": 3.14, # Mean squared error. |
| 296 | "rSquared": 3.14, # R^2 score. |
| 297 | "medianAbsoluteError": 3.14, # Median absolute error. |
| 298 | "meanSquaredLogError": 3.14, # Mean squared log error. |
| 299 | "meanAbsoluteError": 3.14, # Mean absolute error. |
| 300 | }, |
| 301 | "rankingMetrics": { # Evaluation metrics used by weighted-ALS models specified by # [Alpha] Populated for implicit feedback type matrix factorization |
| 302 | # models. |
| 303 | # feedback_type=implicit. |
| 304 | "meanAveragePrecision": 3.14, # Calculates a precision per user for all the items by ranking them and |
| 305 | # then averages all the precisions across all the users. |
| 306 | "normalizedDiscountedCumulativeGain": 3.14, # A metric to determine the goodness of a ranking calculated from the |
| 307 | # predicted confidence by comparing it to an ideal rank measured by the |
| 308 | # original ratings. |
| 309 | "averageRank": 3.14, # Determines the goodness of a ranking by computing the percentile rank |
| 310 | # from the predicted confidence and dividing it by the original rank. |
| 311 | "meanSquaredError": 3.14, # Similar to the mean squared error computed in regression and explicit |
| 312 | # recommendation models except instead of computing the rating directly, |
| 313 | # the output from evaluate is computed against a preference which is 1 or 0 |
| 314 | # depending on if the rating exists or not. |
| 315 | }, |
| 316 | "multiClassClassificationMetrics": { # Evaluation metrics for multi-class classification/classifier models. # Populated for multi-class classification/classifier models. |
| 317 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 318 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 319 | # macro-averaged, the metrics are calculated for each label and then an |
| 320 | # unweighted average is taken of those values. When micro-averaged, the |
| 321 | # metric is calculated globally by counting the total number of correctly |
| 322 | # predicted rows. |
| 323 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 324 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 325 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 326 | # classification models this is the positive class threshold. |
| 327 | # For multi-class classfication models this is the confidence |
| 328 | # threshold. |
| 329 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 330 | # metric. |
| 331 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 332 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 333 | # this is a macro-averaged metric. |
| 334 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 335 | # positive actual labels. For multiclass this is a macro-averaged |
| 336 | # metric treating each class as a binary classifier. |
| 337 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 338 | # multiclass this is a micro-averaged metric. |
| 339 | }, |
| 340 | "confusionMatrixList": [ # Confusion matrix at different thresholds. |
| 341 | { # Confusion matrix for multi-class classification models. |
| 342 | "confidenceThreshold": 3.14, # Confidence threshold used when computing the entries of the |
| 343 | # confusion matrix. |
| 344 | "rows": [ # One row per actual label. |
| 345 | { # A single row in the confusion matrix. |
| 346 | "entries": [ # Info describing predicted label distribution. |
| 347 | { # A single entry in the confusion matrix. |
| 348 | "predictedLabel": "A String", # The predicted label. For confidence_threshold > 0, we will |
| 349 | # also add an entry indicating the number of items under the |
| 350 | # confidence threshold. |
| 351 | "itemCount": "A String", # Number of items being predicted as this label. |
| 352 | }, |
| 353 | ], |
| 354 | "actualLabel": "A String", # The original label of this row. |
| 355 | }, |
| 356 | ], |
| 357 | }, |
| 358 | ], |
| 359 | }, |
| 360 | "clusteringMetrics": { # Evaluation metrics for clustering models. # Populated for clustering models. |
| 361 | "meanSquaredDistance": 3.14, # Mean of squared distances between each sample to its cluster centroid. |
| 362 | "daviesBouldinIndex": 3.14, # Davies-Bouldin index. |
| 363 | "clusters": [ # [Beta] Information for all clusters. |
| 364 | { # Message containing the information about one cluster. |
| 365 | "count": "A String", # Count of training data rows that were assigned to this cluster. |
| 366 | "featureValues": [ # Values of highly variant features for this cluster. |
| 367 | { # Representative value of a single feature within the cluster. |
| 368 | "numericalValue": 3.14, # The numerical feature value. This is the centroid value for this |
| 369 | # feature. |
| 370 | "featureColumn": "A String", # The feature column name. |
| 371 | "categoricalValue": { # Representative value of a categorical feature. # The categorical feature value. |
| 372 | "categoryCounts": [ # Counts of all categories for the categorical feature. If there are |
| 373 | # more than ten categories, we return top ten (by count) and return |
| 374 | # one more CategoryCount with category "_OTHER_" and count as |
| 375 | # aggregate counts of remaining categories. |
| 376 | { # Represents the count of a single category within the cluster. |
| 377 | "category": "A String", # The name of category. |
| 378 | "count": "A String", # The count of training samples matching the category within the |
| 379 | # cluster. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 380 | }, |
| 381 | ], |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 382 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 383 | }, |
| 384 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 385 | "centroidId": "A String", # Centroid id. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 386 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 387 | ], |
| 388 | }, |
| 389 | }, |
| 390 | "trainingOptions": { # Options that were used for this training run, includes |
| 391 | # user specified and default options that were used. |
| 392 | "dropout": 3.14, # Dropout probability for dnn models. |
| 393 | "learnRate": 3.14, # Learning rate in training. Used only for iterative training algorithms. |
| 394 | "labelClassWeights": { # Weights associated with each label class, for rebalancing the |
| 395 | # training data. Only applicable for classification models. |
| 396 | "a_key": 3.14, |
| 397 | }, |
| 398 | "subsample": 3.14, # Subsample fraction of the training data to grow tree to prevent |
| 399 | # overfitting for boosted tree models. |
| 400 | "earlyStop": True or False, # Whether to stop early when the loss doesn't improve significantly |
| 401 | # any more (compared to min_relative_progress). Used only for iterative |
| 402 | # training algorithms. |
| 403 | "dataSplitEvalFraction": 3.14, # The fraction of evaluation data over the whole input data. The rest |
| 404 | # of data will be used as training data. The format should be double. |
| 405 | # Accurate to two decimal places. |
| 406 | # Default value is 0.2. |
| 407 | "initialLearnRate": 3.14, # Specifies the initial learning rate for the line search learn rate |
| 408 | # strategy. |
| 409 | "itemColumn": "A String", # Item column specified for matrix factorization models. |
| 410 | "inputLabelColumns": [ # Name of input label columns in training data. |
| 411 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 412 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 413 | "warmStart": True or False, # Whether to train a model from the last checkpoint. |
| 414 | "learnRateStrategy": "A String", # The strategy to determine learn rate for the current iteration. |
| 415 | "numFactors": "A String", # Num factors specified for matrix factorization models. |
| 416 | "lossType": "A String", # Type of loss function used during training run. |
| 417 | "hiddenUnits": [ # Hidden units for dnn models. |
| 418 | "A String", |
| 419 | ], |
| 420 | "kmeansInitializationMethod": "A String", # The method used to initialize the centroids for kmeans algorithm. |
| 421 | "l1Regularization": 3.14, # L1 regularization coefficient. |
| 422 | "distanceType": "A String", # Distance type for clustering models. |
| 423 | "walsAlpha": 3.14, # Hyperparameter for matrix factoration when implicit feedback type is |
| 424 | # specified. |
| 425 | "feedbackType": "A String", # Feedback type that specifies which algorithm to run for matrix |
| 426 | # factorization. |
| 427 | "optimizationStrategy": "A String", # Optimization strategy for training linear regression models. |
| 428 | "dataSplitColumn": "A String", # The column to split data with. This column won't be used as a |
| 429 | # feature. |
| 430 | # 1. When data_split_method is CUSTOM, the corresponding column should |
| 431 | # be boolean. The rows with true value tag are eval data, and the false |
| 432 | # are training data. |
| 433 | # 2. When data_split_method is SEQ, the first DATA_SPLIT_EVAL_FRACTION |
| 434 | # rows (from smallest to largest) in the corresponding column are used |
| 435 | # as training data, and the rest are eval data. It respects the order |
| 436 | # in Orderable data types: |
| 437 | # https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#data-type-properties |
| 438 | "maxIterations": "A String", # The maximum number of iterations in training. Used only for iterative |
| 439 | # training algorithms. |
| 440 | "userColumn": "A String", # User column specified for matrix factorization models. |
| 441 | "maxTreeDepth": "A String", # Maximum depth of a tree for boosted tree models. |
| 442 | "l2Regularization": 3.14, # L2 regularization coefficient. |
| 443 | "modelUri": "A String", # [Beta] Google Cloud Storage URI from which the model was imported. Only |
| 444 | # applicable for imported models. |
| 445 | "batchSize": "A String", # Batch size for dnn models. |
| 446 | "minRelativeProgress": 3.14, # When early_stop is true, stops training when accuracy improvement is |
| 447 | # less than 'min_relative_progress'. Used only for iterative training |
| 448 | # algorithms. |
| 449 | "kmeansInitializationColumn": "A String", # The column used to provide the initial centroids for kmeans algorithm |
| 450 | # when kmeans_initialization_method is CUSTOM. |
| 451 | "numClusters": "A String", # Number of clusters for clustering models. |
| 452 | "dataSplitMethod": "A String", # The data split type for training and evaluation, e.g. RANDOM. |
| 453 | "minSplitLoss": 3.14, # Minimum split loss for boosted tree models. |
| 454 | }, |
| 455 | "dataSplitResult": { # Data split result. This contains references to the training and evaluation # Data split result of the training run. Only set when the input data is |
| 456 | # actually split. |
| 457 | # data tables that were used to train the model. |
| 458 | "trainingTable": { # Table reference of the training data after split. |
| 459 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 460 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 461 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| 462 | }, |
| 463 | "evaluationTable": { # Table reference of the evaluation data after split. |
| 464 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 465 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 466 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 467 | }, |
| 468 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 469 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 470 | ], |
| 471 | "modelReference": { # Required. Unique identifier for this model. |
| 472 | "datasetId": "A String", # [Required] The ID of the dataset containing this model. |
| 473 | "modelId": "A String", # [Required] The ID of the model. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 474 | "projectId": "A String", # [Required] The ID of the project containing this model. |
| 475 | }, |
| 476 | "description": "A String", # Optional. A user-friendly description of this model. |
| 477 | "etag": "A String", # Output only. A hash of this resource. |
| 478 | "creationTime": "A String", # Output only. The time when this model was created, in millisecs since the epoch. |
| 479 | "encryptionConfiguration": { # Custom encryption configuration (e.g., Cloud KMS keys). This shows the |
| 480 | # encryption configuration of the model data while stored in BigQuery |
| 481 | # storage. This field can be used with PatchModel to update encryption key |
| 482 | # for an already encrypted model. |
| 483 | "kmsKeyName": "A String", # [Optional] Describes the Cloud KMS encryption key that will be used to protect destination BigQuery table. The BigQuery Service Account associated with your project requires access to this encryption key. |
| 484 | }, |
| 485 | }</pre> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 486 | </div> |
| 487 | |
| 488 | <div class="method"> |
| 489 | <code class="details" id="list">list(projectId, datasetId, pageToken=None, maxResults=None)</code> |
| 490 | <pre>Lists all models in the specified dataset. Requires the READER dataset |
| 491 | role. |
| 492 | |
| 493 | Args: |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 494 | projectId: string, Required. Project ID of the models to list. (required) |
| 495 | datasetId: string, Required. Dataset ID of the models to list. (required) |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 496 | pageToken: string, Page token, returned by a previous call to request the next page of |
| 497 | results |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 498 | maxResults: integer, The maximum number of results to return in a single response page. |
| 499 | Leverage the page tokens to iterate through the entire collection. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 500 | |
| 501 | Returns: |
| 502 | An object of the form: |
| 503 | |
| 504 | { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 505 | "nextPageToken": "A String", # A token to request the next page of results. |
| 506 | "models": [ # Models in the requested dataset. Only the following fields are populated: |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 507 | # model_reference, model_type, creation_time, last_modified_time and |
| 508 | # labels. |
| 509 | { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 510 | "location": "A String", # Output only. The geographic location where the model resides. This value |
| 511 | # is inherited from the dataset. |
| 512 | "friendlyName": "A String", # Optional. A descriptive name for this model. |
| 513 | "lastModifiedTime": "A String", # Output only. The time when this model was last modified, in millisecs since the epoch. |
| 514 | "labels": { # The labels associated with this model. You can use these to organize |
| 515 | # and group your models. Label keys and values can be no longer |
| 516 | # than 63 characters, can only contain lowercase letters, numeric |
| 517 | # characters, underscores and dashes. International characters are allowed. |
| 518 | # Label values are optional. Label keys must start with a letter and each |
| 519 | # label in the list must have a different key. |
| 520 | "a_key": "A String", |
| 521 | }, |
| 522 | "labelColumns": [ # Output only. Label columns that were used to train this model. |
| 523 | # The output of the model will have a "predicted_" prefix to these columns. |
| 524 | { # A field or a column. |
| 525 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 526 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 527 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 528 | # in this case the output parameter does not have this "type" field). |
| 529 | # Examples: |
| 530 | # INT64: {type_kind="INT64"} |
| 531 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 532 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 533 | # {type_kind="STRUCT", |
| 534 | # struct_type={fields=[ |
| 535 | # {name="x", type={type_kind="STRING"}}, |
| 536 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 537 | # ]}} |
| 538 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 539 | "fields": [ |
| 540 | # Object with schema name: StandardSqlField |
| 541 | ], |
| 542 | }, |
| 543 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 544 | "typeKind": "A String", # Required. The top level type of this field. |
| 545 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 546 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 547 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 548 | ], |
| 549 | "modelType": "A String", # Output only. Type of the model resource. |
| 550 | "featureColumns": [ # Output only. Input feature columns that were used to train this model. |
| 551 | { # A field or a column. |
| 552 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 553 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 554 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 555 | # in this case the output parameter does not have this "type" field). |
| 556 | # Examples: |
| 557 | # INT64: {type_kind="INT64"} |
| 558 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 559 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 560 | # {type_kind="STRUCT", |
| 561 | # struct_type={fields=[ |
| 562 | # {name="x", type={type_kind="STRING"}}, |
| 563 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 564 | # ]}} |
| 565 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 566 | "fields": [ |
| 567 | # Object with schema name: StandardSqlField |
| 568 | ], |
| 569 | }, |
| 570 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 571 | "typeKind": "A String", # Required. The top level type of this field. |
| 572 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 573 | }, |
| 574 | }, |
| 575 | ], |
| 576 | "expirationTime": "A String", # Optional. The time when this model expires, in milliseconds since the epoch. |
| 577 | # If not present, the model will persist indefinitely. Expired models |
| 578 | # will be deleted and their storage reclaimed. The defaultTableExpirationMs |
| 579 | # property of the encapsulating dataset can be used to set a default |
| 580 | # expirationTime on newly created models. |
| 581 | "trainingRuns": [ # Output only. Information for all training runs in increasing order of start_time. |
| 582 | { # Information about a single training query run for the model. |
| 583 | "startTime": "A String", # The start time of this training run. |
| 584 | "results": [ # Output of each iteration run, results.size() <= max_iterations. |
| 585 | { # Information about a single iteration of the training run. |
| 586 | "trainingLoss": 3.14, # Loss computed on the training data at the end of iteration. |
| 587 | "evalLoss": 3.14, # Loss computed on the eval data at the end of iteration. |
| 588 | "index": 42, # Index of the iteration, 0 based. |
| 589 | "learnRate": 3.14, # Learn rate used for this iteration. |
| 590 | "durationMs": "A String", # Time taken to run the iteration in milliseconds. |
| 591 | "arimaResult": { # (Auto-)arima fitting result. Wrap everything in ArimaResult for easier |
| 592 | # refactoring if we want to use model-specific iteration results. |
| 593 | "arimaModelInfo": [ # This message is repeated because there are multiple arima models |
| 594 | # fitted in auto-arima. For non-auto-arima model, its size is one. |
| 595 | { # Arima model information. |
| 596 | "arimaFittingMetrics": { # ARIMA model fitting metrics. # Arima fitting metrics. |
| 597 | "aic": 3.14, # AIC. |
| 598 | "logLikelihood": 3.14, # Log-likelihood. |
| 599 | "variance": 3.14, # Variance. |
| 600 | }, |
| 601 | "timeSeriesId": "A String", # The id to indicate different time series. |
| 602 | "arimaCoefficients": { # Arima coefficients. # Arima coefficients. |
| 603 | "movingAverageCoefficients": [ # Moving-average coefficients, an array of double. |
| 604 | 3.14, |
| 605 | ], |
| 606 | "autoRegressiveCoefficients": [ # Auto-regressive coefficients, an array of double. |
| 607 | 3.14, |
| 608 | ], |
| 609 | "interceptCoefficient": 3.14, # Intercept coefficient, just a double not an array. |
| 610 | }, |
| 611 | "hasDrift": True or False, # Whether Arima model fitted with drift or not. It is always false |
| 612 | # when d is not 1. |
| 613 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported |
| 614 | # for one time series. |
| 615 | "A String", |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 616 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 617 | "nonSeasonalOrder": { # Arima order, can be used for both non-seasonal and seasonal parts. # Non-seasonal order. |
| 618 | "q": "A String", # Order of the moving-average part. |
| 619 | "d": "A String", # Order of the differencing part. |
| 620 | "p": "A String", # Order of the autoregressive part. |
| 621 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 622 | }, |
| 623 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 624 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported for |
| 625 | # one time series. |
| 626 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 627 | ], |
| 628 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 629 | "clusterInfos": [ # Information about top clusters for clustering models. |
| 630 | { # Information about a single cluster for clustering model. |
| 631 | "clusterSize": "A String", # Cluster size, the total number of points assigned to the cluster. |
| 632 | "centroidId": "A String", # Centroid id. |
| 633 | "clusterRadius": 3.14, # Cluster radius, the average distance from centroid |
| 634 | # to each point assigned to the cluster. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 635 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 636 | ], |
| 637 | }, |
| 638 | ], |
| 639 | "evaluationMetrics": { # Evaluation metrics of a model. These are either computed on all training # The evaluation metrics over training/eval data that were computed at the |
| 640 | # end of training. |
| 641 | # data or just the eval data based on whether eval data was used during |
| 642 | # training. These are not present for imported models. |
| 643 | "binaryClassificationMetrics": { # Evaluation metrics for binary classification/classifier models. # Populated for binary classification/classifier models. |
| 644 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 645 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 646 | # macro-averaged, the metrics are calculated for each label and then an |
| 647 | # unweighted average is taken of those values. When micro-averaged, the |
| 648 | # metric is calculated globally by counting the total number of correctly |
| 649 | # predicted rows. |
| 650 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 651 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 652 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 653 | # classification models this is the positive class threshold. |
| 654 | # For multi-class classfication models this is the confidence |
| 655 | # threshold. |
| 656 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 657 | # metric. |
| 658 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 659 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 660 | # this is a macro-averaged metric. |
| 661 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 662 | # positive actual labels. For multiclass this is a macro-averaged |
| 663 | # metric treating each class as a binary classifier. |
| 664 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 665 | # multiclass this is a micro-averaged metric. |
| 666 | }, |
| 667 | "negativeLabel": "A String", # Label representing the negative class. |
| 668 | "positiveLabel": "A String", # Label representing the positive class. |
| 669 | "binaryConfusionMatrixList": [ # Binary confusion matrix at multiple thresholds. |
| 670 | { # Confusion matrix for binary classification models. |
| 671 | "falseNegatives": "A String", # Number of false samples predicted as false. |
| 672 | "falsePositives": "A String", # Number of false samples predicted as true. |
| 673 | "trueNegatives": "A String", # Number of true samples predicted as false. |
| 674 | "f1Score": 3.14, # The equally weighted average of recall and precision. |
| 675 | "precision": 3.14, # The fraction of actual positive predictions that had positive actual |
| 676 | # labels. |
| 677 | "positiveClassThreshold": 3.14, # Threshold value used when computing each of the following metric. |
| 678 | "accuracy": 3.14, # The fraction of predictions given the correct label. |
| 679 | "truePositives": "A String", # Number of true samples predicted as true. |
| 680 | "recall": 3.14, # The fraction of actual positive labels that were given a positive |
| 681 | # prediction. |
| 682 | }, |
| 683 | ], |
| 684 | }, |
| 685 | "regressionMetrics": { # Evaluation metrics for regression and explicit feedback type matrix # Populated for regression models and explicit feedback type matrix |
| 686 | # factorization models. |
| 687 | # factorization models. |
| 688 | "meanSquaredError": 3.14, # Mean squared error. |
| 689 | "rSquared": 3.14, # R^2 score. |
| 690 | "medianAbsoluteError": 3.14, # Median absolute error. |
| 691 | "meanSquaredLogError": 3.14, # Mean squared log error. |
| 692 | "meanAbsoluteError": 3.14, # Mean absolute error. |
| 693 | }, |
| 694 | "rankingMetrics": { # Evaluation metrics used by weighted-ALS models specified by # [Alpha] Populated for implicit feedback type matrix factorization |
| 695 | # models. |
| 696 | # feedback_type=implicit. |
| 697 | "meanAveragePrecision": 3.14, # Calculates a precision per user for all the items by ranking them and |
| 698 | # then averages all the precisions across all the users. |
| 699 | "normalizedDiscountedCumulativeGain": 3.14, # A metric to determine the goodness of a ranking calculated from the |
| 700 | # predicted confidence by comparing it to an ideal rank measured by the |
| 701 | # original ratings. |
| 702 | "averageRank": 3.14, # Determines the goodness of a ranking by computing the percentile rank |
| 703 | # from the predicted confidence and dividing it by the original rank. |
| 704 | "meanSquaredError": 3.14, # Similar to the mean squared error computed in regression and explicit |
| 705 | # recommendation models except instead of computing the rating directly, |
| 706 | # the output from evaluate is computed against a preference which is 1 or 0 |
| 707 | # depending on if the rating exists or not. |
| 708 | }, |
| 709 | "multiClassClassificationMetrics": { # Evaluation metrics for multi-class classification/classifier models. # Populated for multi-class classification/classifier models. |
| 710 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 711 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 712 | # macro-averaged, the metrics are calculated for each label and then an |
| 713 | # unweighted average is taken of those values. When micro-averaged, the |
| 714 | # metric is calculated globally by counting the total number of correctly |
| 715 | # predicted rows. |
| 716 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 717 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 718 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 719 | # classification models this is the positive class threshold. |
| 720 | # For multi-class classfication models this is the confidence |
| 721 | # threshold. |
| 722 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 723 | # metric. |
| 724 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 725 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 726 | # this is a macro-averaged metric. |
| 727 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 728 | # positive actual labels. For multiclass this is a macro-averaged |
| 729 | # metric treating each class as a binary classifier. |
| 730 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 731 | # multiclass this is a micro-averaged metric. |
| 732 | }, |
| 733 | "confusionMatrixList": [ # Confusion matrix at different thresholds. |
| 734 | { # Confusion matrix for multi-class classification models. |
| 735 | "confidenceThreshold": 3.14, # Confidence threshold used when computing the entries of the |
| 736 | # confusion matrix. |
| 737 | "rows": [ # One row per actual label. |
| 738 | { # A single row in the confusion matrix. |
| 739 | "entries": [ # Info describing predicted label distribution. |
| 740 | { # A single entry in the confusion matrix. |
| 741 | "predictedLabel": "A String", # The predicted label. For confidence_threshold > 0, we will |
| 742 | # also add an entry indicating the number of items under the |
| 743 | # confidence threshold. |
| 744 | "itemCount": "A String", # Number of items being predicted as this label. |
| 745 | }, |
| 746 | ], |
| 747 | "actualLabel": "A String", # The original label of this row. |
| 748 | }, |
| 749 | ], |
| 750 | }, |
| 751 | ], |
| 752 | }, |
| 753 | "clusteringMetrics": { # Evaluation metrics for clustering models. # Populated for clustering models. |
| 754 | "meanSquaredDistance": 3.14, # Mean of squared distances between each sample to its cluster centroid. |
| 755 | "daviesBouldinIndex": 3.14, # Davies-Bouldin index. |
| 756 | "clusters": [ # [Beta] Information for all clusters. |
| 757 | { # Message containing the information about one cluster. |
| 758 | "count": "A String", # Count of training data rows that were assigned to this cluster. |
| 759 | "featureValues": [ # Values of highly variant features for this cluster. |
| 760 | { # Representative value of a single feature within the cluster. |
| 761 | "numericalValue": 3.14, # The numerical feature value. This is the centroid value for this |
| 762 | # feature. |
| 763 | "featureColumn": "A String", # The feature column name. |
| 764 | "categoricalValue": { # Representative value of a categorical feature. # The categorical feature value. |
| 765 | "categoryCounts": [ # Counts of all categories for the categorical feature. If there are |
| 766 | # more than ten categories, we return top ten (by count) and return |
| 767 | # one more CategoryCount with category "_OTHER_" and count as |
| 768 | # aggregate counts of remaining categories. |
| 769 | { # Represents the count of a single category within the cluster. |
| 770 | "category": "A String", # The name of category. |
| 771 | "count": "A String", # The count of training samples matching the category within the |
| 772 | # cluster. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 773 | }, |
| 774 | ], |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 775 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 776 | }, |
| 777 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 778 | "centroidId": "A String", # Centroid id. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 779 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 780 | ], |
| 781 | }, |
| 782 | }, |
| 783 | "trainingOptions": { # Options that were used for this training run, includes |
| 784 | # user specified and default options that were used. |
| 785 | "dropout": 3.14, # Dropout probability for dnn models. |
| 786 | "learnRate": 3.14, # Learning rate in training. Used only for iterative training algorithms. |
| 787 | "labelClassWeights": { # Weights associated with each label class, for rebalancing the |
| 788 | # training data. Only applicable for classification models. |
| 789 | "a_key": 3.14, |
| 790 | }, |
| 791 | "subsample": 3.14, # Subsample fraction of the training data to grow tree to prevent |
| 792 | # overfitting for boosted tree models. |
| 793 | "earlyStop": True or False, # Whether to stop early when the loss doesn't improve significantly |
| 794 | # any more (compared to min_relative_progress). Used only for iterative |
| 795 | # training algorithms. |
| 796 | "dataSplitEvalFraction": 3.14, # The fraction of evaluation data over the whole input data. The rest |
| 797 | # of data will be used as training data. The format should be double. |
| 798 | # Accurate to two decimal places. |
| 799 | # Default value is 0.2. |
| 800 | "initialLearnRate": 3.14, # Specifies the initial learning rate for the line search learn rate |
| 801 | # strategy. |
| 802 | "itemColumn": "A String", # Item column specified for matrix factorization models. |
| 803 | "inputLabelColumns": [ # Name of input label columns in training data. |
| 804 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 805 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 806 | "warmStart": True or False, # Whether to train a model from the last checkpoint. |
| 807 | "learnRateStrategy": "A String", # The strategy to determine learn rate for the current iteration. |
| 808 | "numFactors": "A String", # Num factors specified for matrix factorization models. |
| 809 | "lossType": "A String", # Type of loss function used during training run. |
| 810 | "hiddenUnits": [ # Hidden units for dnn models. |
| 811 | "A String", |
| 812 | ], |
| 813 | "kmeansInitializationMethod": "A String", # The method used to initialize the centroids for kmeans algorithm. |
| 814 | "l1Regularization": 3.14, # L1 regularization coefficient. |
| 815 | "distanceType": "A String", # Distance type for clustering models. |
| 816 | "walsAlpha": 3.14, # Hyperparameter for matrix factoration when implicit feedback type is |
| 817 | # specified. |
| 818 | "feedbackType": "A String", # Feedback type that specifies which algorithm to run for matrix |
| 819 | # factorization. |
| 820 | "optimizationStrategy": "A String", # Optimization strategy for training linear regression models. |
| 821 | "dataSplitColumn": "A String", # The column to split data with. This column won't be used as a |
| 822 | # feature. |
| 823 | # 1. When data_split_method is CUSTOM, the corresponding column should |
| 824 | # be boolean. The rows with true value tag are eval data, and the false |
| 825 | # are training data. |
| 826 | # 2. When data_split_method is SEQ, the first DATA_SPLIT_EVAL_FRACTION |
| 827 | # rows (from smallest to largest) in the corresponding column are used |
| 828 | # as training data, and the rest are eval data. It respects the order |
| 829 | # in Orderable data types: |
| 830 | # https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#data-type-properties |
| 831 | "maxIterations": "A String", # The maximum number of iterations in training. Used only for iterative |
| 832 | # training algorithms. |
| 833 | "userColumn": "A String", # User column specified for matrix factorization models. |
| 834 | "maxTreeDepth": "A String", # Maximum depth of a tree for boosted tree models. |
| 835 | "l2Regularization": 3.14, # L2 regularization coefficient. |
| 836 | "modelUri": "A String", # [Beta] Google Cloud Storage URI from which the model was imported. Only |
| 837 | # applicable for imported models. |
| 838 | "batchSize": "A String", # Batch size for dnn models. |
| 839 | "minRelativeProgress": 3.14, # When early_stop is true, stops training when accuracy improvement is |
| 840 | # less than 'min_relative_progress'. Used only for iterative training |
| 841 | # algorithms. |
| 842 | "kmeansInitializationColumn": "A String", # The column used to provide the initial centroids for kmeans algorithm |
| 843 | # when kmeans_initialization_method is CUSTOM. |
| 844 | "numClusters": "A String", # Number of clusters for clustering models. |
| 845 | "dataSplitMethod": "A String", # The data split type for training and evaluation, e.g. RANDOM. |
| 846 | "minSplitLoss": 3.14, # Minimum split loss for boosted tree models. |
| 847 | }, |
| 848 | "dataSplitResult": { # Data split result. This contains references to the training and evaluation # Data split result of the training run. Only set when the input data is |
| 849 | # actually split. |
| 850 | # data tables that were used to train the model. |
| 851 | "trainingTable": { # Table reference of the training data after split. |
| 852 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 853 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 854 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| 855 | }, |
| 856 | "evaluationTable": { # Table reference of the evaluation data after split. |
| 857 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 858 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 859 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 860 | }, |
| 861 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 862 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 863 | ], |
| 864 | "modelReference": { # Required. Unique identifier for this model. |
| 865 | "datasetId": "A String", # [Required] The ID of the dataset containing this model. |
| 866 | "modelId": "A String", # [Required] The ID of the model. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 867 | "projectId": "A String", # [Required] The ID of the project containing this model. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 868 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 869 | "description": "A String", # Optional. A user-friendly description of this model. |
| 870 | "etag": "A String", # Output only. A hash of this resource. |
| 871 | "creationTime": "A String", # Output only. The time when this model was created, in millisecs since the epoch. |
| 872 | "encryptionConfiguration": { # Custom encryption configuration (e.g., Cloud KMS keys). This shows the |
| 873 | # encryption configuration of the model data while stored in BigQuery |
| 874 | # storage. This field can be used with PatchModel to update encryption key |
| 875 | # for an already encrypted model. |
| 876 | "kmsKeyName": "A String", # [Optional] Describes the Cloud KMS encryption key that will be used to protect destination BigQuery table. The BigQuery Service Account associated with your project requires access to this encryption key. |
| 877 | }, |
| 878 | }, |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 879 | ], |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 880 | }</pre> |
| 881 | </div> |
| 882 | |
| 883 | <div class="method"> |
| 884 | <code class="details" id="list_next">list_next(previous_request, previous_response)</code> |
| 885 | <pre>Retrieves the next page of results. |
| 886 | |
| 887 | Args: |
| 888 | previous_request: The request for the previous page. (required) |
| 889 | previous_response: The response from the request for the previous page. (required) |
| 890 | |
| 891 | Returns: |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 892 | A request object that you can call 'execute()' on to request the next |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 893 | page. Returns None if there are no more items in the collection. |
| 894 | </pre> |
| 895 | </div> |
| 896 | |
| 897 | <div class="method"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 898 | <code class="details" id="patch">patch(projectId, datasetId, modelId, body=None)</code> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 899 | <pre>Patch specific fields in the specified model. |
| 900 | |
| 901 | Args: |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 902 | projectId: string, Required. Project ID of the model to patch. (required) |
| 903 | datasetId: string, Required. Dataset ID of the model to patch. (required) |
| 904 | modelId: string, Required. Model ID of the model to patch. (required) |
| 905 | body: object, The request body. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 906 | The object takes the form of: |
| 907 | |
| 908 | { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 909 | "location": "A String", # Output only. The geographic location where the model resides. This value |
| 910 | # is inherited from the dataset. |
| 911 | "friendlyName": "A String", # Optional. A descriptive name for this model. |
| 912 | "lastModifiedTime": "A String", # Output only. The time when this model was last modified, in millisecs since the epoch. |
| 913 | "labels": { # The labels associated with this model. You can use these to organize |
| 914 | # and group your models. Label keys and values can be no longer |
| 915 | # than 63 characters, can only contain lowercase letters, numeric |
| 916 | # characters, underscores and dashes. International characters are allowed. |
| 917 | # Label values are optional. Label keys must start with a letter and each |
| 918 | # label in the list must have a different key. |
| 919 | "a_key": "A String", |
| 920 | }, |
| 921 | "labelColumns": [ # Output only. Label columns that were used to train this model. |
| 922 | # The output of the model will have a "predicted_" prefix to these columns. |
| 923 | { # A field or a column. |
| 924 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 925 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 926 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 927 | # in this case the output parameter does not have this "type" field). |
| 928 | # Examples: |
| 929 | # INT64: {type_kind="INT64"} |
| 930 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 931 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 932 | # {type_kind="STRUCT", |
| 933 | # struct_type={fields=[ |
| 934 | # {name="x", type={type_kind="STRING"}}, |
| 935 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 936 | # ]}} |
| 937 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 938 | "fields": [ |
| 939 | # Object with schema name: StandardSqlField |
| 940 | ], |
| 941 | }, |
| 942 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 943 | "typeKind": "A String", # Required. The top level type of this field. |
| 944 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 945 | }, |
| 946 | }, |
| 947 | ], |
| 948 | "modelType": "A String", # Output only. Type of the model resource. |
| 949 | "featureColumns": [ # Output only. Input feature columns that were used to train this model. |
| 950 | { # A field or a column. |
| 951 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 952 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 953 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 954 | # in this case the output parameter does not have this "type" field). |
| 955 | # Examples: |
| 956 | # INT64: {type_kind="INT64"} |
| 957 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 958 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 959 | # {type_kind="STRUCT", |
| 960 | # struct_type={fields=[ |
| 961 | # {name="x", type={type_kind="STRING"}}, |
| 962 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 963 | # ]}} |
| 964 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 965 | "fields": [ |
| 966 | # Object with schema name: StandardSqlField |
| 967 | ], |
| 968 | }, |
| 969 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 970 | "typeKind": "A String", # Required. The top level type of this field. |
| 971 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 972 | }, |
| 973 | }, |
| 974 | ], |
| 975 | "expirationTime": "A String", # Optional. The time when this model expires, in milliseconds since the epoch. |
| 976 | # If not present, the model will persist indefinitely. Expired models |
| 977 | # will be deleted and their storage reclaimed. The defaultTableExpirationMs |
| 978 | # property of the encapsulating dataset can be used to set a default |
| 979 | # expirationTime on newly created models. |
| 980 | "trainingRuns": [ # Output only. Information for all training runs in increasing order of start_time. |
| 981 | { # Information about a single training query run for the model. |
| 982 | "startTime": "A String", # The start time of this training run. |
| 983 | "results": [ # Output of each iteration run, results.size() <= max_iterations. |
| 984 | { # Information about a single iteration of the training run. |
| 985 | "trainingLoss": 3.14, # Loss computed on the training data at the end of iteration. |
| 986 | "evalLoss": 3.14, # Loss computed on the eval data at the end of iteration. |
| 987 | "index": 42, # Index of the iteration, 0 based. |
| 988 | "learnRate": 3.14, # Learn rate used for this iteration. |
| 989 | "durationMs": "A String", # Time taken to run the iteration in milliseconds. |
| 990 | "arimaResult": { # (Auto-)arima fitting result. Wrap everything in ArimaResult for easier |
| 991 | # refactoring if we want to use model-specific iteration results. |
| 992 | "arimaModelInfo": [ # This message is repeated because there are multiple arima models |
| 993 | # fitted in auto-arima. For non-auto-arima model, its size is one. |
| 994 | { # Arima model information. |
| 995 | "arimaFittingMetrics": { # ARIMA model fitting metrics. # Arima fitting metrics. |
| 996 | "aic": 3.14, # AIC. |
| 997 | "logLikelihood": 3.14, # Log-likelihood. |
| 998 | "variance": 3.14, # Variance. |
| 999 | }, |
| 1000 | "timeSeriesId": "A String", # The id to indicate different time series. |
| 1001 | "arimaCoefficients": { # Arima coefficients. # Arima coefficients. |
| 1002 | "movingAverageCoefficients": [ # Moving-average coefficients, an array of double. |
| 1003 | 3.14, |
| 1004 | ], |
| 1005 | "autoRegressiveCoefficients": [ # Auto-regressive coefficients, an array of double. |
| 1006 | 3.14, |
| 1007 | ], |
| 1008 | "interceptCoefficient": 3.14, # Intercept coefficient, just a double not an array. |
| 1009 | }, |
| 1010 | "hasDrift": True or False, # Whether Arima model fitted with drift or not. It is always false |
| 1011 | # when d is not 1. |
| 1012 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported |
| 1013 | # for one time series. |
| 1014 | "A String", |
| 1015 | ], |
| 1016 | "nonSeasonalOrder": { # Arima order, can be used for both non-seasonal and seasonal parts. # Non-seasonal order. |
| 1017 | "q": "A String", # Order of the moving-average part. |
| 1018 | "d": "A String", # Order of the differencing part. |
| 1019 | "p": "A String", # Order of the autoregressive part. |
| 1020 | }, |
| 1021 | }, |
| 1022 | ], |
| 1023 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported for |
| 1024 | # one time series. |
| 1025 | "A String", |
| 1026 | ], |
| 1027 | }, |
| 1028 | "clusterInfos": [ # Information about top clusters for clustering models. |
| 1029 | { # Information about a single cluster for clustering model. |
| 1030 | "clusterSize": "A String", # Cluster size, the total number of points assigned to the cluster. |
| 1031 | "centroidId": "A String", # Centroid id. |
| 1032 | "clusterRadius": 3.14, # Cluster radius, the average distance from centroid |
| 1033 | # to each point assigned to the cluster. |
| 1034 | }, |
| 1035 | ], |
| 1036 | }, |
| 1037 | ], |
| 1038 | "evaluationMetrics": { # Evaluation metrics of a model. These are either computed on all training # The evaluation metrics over training/eval data that were computed at the |
| 1039 | # end of training. |
| 1040 | # data or just the eval data based on whether eval data was used during |
| 1041 | # training. These are not present for imported models. |
| 1042 | "binaryClassificationMetrics": { # Evaluation metrics for binary classification/classifier models. # Populated for binary classification/classifier models. |
| 1043 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 1044 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 1045 | # macro-averaged, the metrics are calculated for each label and then an |
| 1046 | # unweighted average is taken of those values. When micro-averaged, the |
| 1047 | # metric is calculated globally by counting the total number of correctly |
| 1048 | # predicted rows. |
| 1049 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 1050 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 1051 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 1052 | # classification models this is the positive class threshold. |
| 1053 | # For multi-class classfication models this is the confidence |
| 1054 | # threshold. |
| 1055 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 1056 | # metric. |
| 1057 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 1058 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 1059 | # this is a macro-averaged metric. |
| 1060 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 1061 | # positive actual labels. For multiclass this is a macro-averaged |
| 1062 | # metric treating each class as a binary classifier. |
| 1063 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 1064 | # multiclass this is a micro-averaged metric. |
| 1065 | }, |
| 1066 | "negativeLabel": "A String", # Label representing the negative class. |
| 1067 | "positiveLabel": "A String", # Label representing the positive class. |
| 1068 | "binaryConfusionMatrixList": [ # Binary confusion matrix at multiple thresholds. |
| 1069 | { # Confusion matrix for binary classification models. |
| 1070 | "falseNegatives": "A String", # Number of false samples predicted as false. |
| 1071 | "falsePositives": "A String", # Number of false samples predicted as true. |
| 1072 | "trueNegatives": "A String", # Number of true samples predicted as false. |
| 1073 | "f1Score": 3.14, # The equally weighted average of recall and precision. |
| 1074 | "precision": 3.14, # The fraction of actual positive predictions that had positive actual |
| 1075 | # labels. |
| 1076 | "positiveClassThreshold": 3.14, # Threshold value used when computing each of the following metric. |
| 1077 | "accuracy": 3.14, # The fraction of predictions given the correct label. |
| 1078 | "truePositives": "A String", # Number of true samples predicted as true. |
| 1079 | "recall": 3.14, # The fraction of actual positive labels that were given a positive |
| 1080 | # prediction. |
| 1081 | }, |
| 1082 | ], |
| 1083 | }, |
| 1084 | "regressionMetrics": { # Evaluation metrics for regression and explicit feedback type matrix # Populated for regression models and explicit feedback type matrix |
| 1085 | # factorization models. |
| 1086 | # factorization models. |
| 1087 | "meanSquaredError": 3.14, # Mean squared error. |
| 1088 | "rSquared": 3.14, # R^2 score. |
| 1089 | "medianAbsoluteError": 3.14, # Median absolute error. |
| 1090 | "meanSquaredLogError": 3.14, # Mean squared log error. |
| 1091 | "meanAbsoluteError": 3.14, # Mean absolute error. |
| 1092 | }, |
| 1093 | "rankingMetrics": { # Evaluation metrics used by weighted-ALS models specified by # [Alpha] Populated for implicit feedback type matrix factorization |
| 1094 | # models. |
| 1095 | # feedback_type=implicit. |
| 1096 | "meanAveragePrecision": 3.14, # Calculates a precision per user for all the items by ranking them and |
| 1097 | # then averages all the precisions across all the users. |
| 1098 | "normalizedDiscountedCumulativeGain": 3.14, # A metric to determine the goodness of a ranking calculated from the |
| 1099 | # predicted confidence by comparing it to an ideal rank measured by the |
| 1100 | # original ratings. |
| 1101 | "averageRank": 3.14, # Determines the goodness of a ranking by computing the percentile rank |
| 1102 | # from the predicted confidence and dividing it by the original rank. |
| 1103 | "meanSquaredError": 3.14, # Similar to the mean squared error computed in regression and explicit |
| 1104 | # recommendation models except instead of computing the rating directly, |
| 1105 | # the output from evaluate is computed against a preference which is 1 or 0 |
| 1106 | # depending on if the rating exists or not. |
| 1107 | }, |
| 1108 | "multiClassClassificationMetrics": { # Evaluation metrics for multi-class classification/classifier models. # Populated for multi-class classification/classifier models. |
| 1109 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 1110 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 1111 | # macro-averaged, the metrics are calculated for each label and then an |
| 1112 | # unweighted average is taken of those values. When micro-averaged, the |
| 1113 | # metric is calculated globally by counting the total number of correctly |
| 1114 | # predicted rows. |
| 1115 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 1116 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 1117 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 1118 | # classification models this is the positive class threshold. |
| 1119 | # For multi-class classfication models this is the confidence |
| 1120 | # threshold. |
| 1121 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 1122 | # metric. |
| 1123 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 1124 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 1125 | # this is a macro-averaged metric. |
| 1126 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 1127 | # positive actual labels. For multiclass this is a macro-averaged |
| 1128 | # metric treating each class as a binary classifier. |
| 1129 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 1130 | # multiclass this is a micro-averaged metric. |
| 1131 | }, |
| 1132 | "confusionMatrixList": [ # Confusion matrix at different thresholds. |
| 1133 | { # Confusion matrix for multi-class classification models. |
| 1134 | "confidenceThreshold": 3.14, # Confidence threshold used when computing the entries of the |
| 1135 | # confusion matrix. |
| 1136 | "rows": [ # One row per actual label. |
| 1137 | { # A single row in the confusion matrix. |
| 1138 | "entries": [ # Info describing predicted label distribution. |
| 1139 | { # A single entry in the confusion matrix. |
| 1140 | "predictedLabel": "A String", # The predicted label. For confidence_threshold > 0, we will |
| 1141 | # also add an entry indicating the number of items under the |
| 1142 | # confidence threshold. |
| 1143 | "itemCount": "A String", # Number of items being predicted as this label. |
| 1144 | }, |
| 1145 | ], |
| 1146 | "actualLabel": "A String", # The original label of this row. |
| 1147 | }, |
| 1148 | ], |
| 1149 | }, |
| 1150 | ], |
| 1151 | }, |
| 1152 | "clusteringMetrics": { # Evaluation metrics for clustering models. # Populated for clustering models. |
| 1153 | "meanSquaredDistance": 3.14, # Mean of squared distances between each sample to its cluster centroid. |
| 1154 | "daviesBouldinIndex": 3.14, # Davies-Bouldin index. |
| 1155 | "clusters": [ # [Beta] Information for all clusters. |
| 1156 | { # Message containing the information about one cluster. |
| 1157 | "count": "A String", # Count of training data rows that were assigned to this cluster. |
| 1158 | "featureValues": [ # Values of highly variant features for this cluster. |
| 1159 | { # Representative value of a single feature within the cluster. |
| 1160 | "numericalValue": 3.14, # The numerical feature value. This is the centroid value for this |
| 1161 | # feature. |
| 1162 | "featureColumn": "A String", # The feature column name. |
| 1163 | "categoricalValue": { # Representative value of a categorical feature. # The categorical feature value. |
| 1164 | "categoryCounts": [ # Counts of all categories for the categorical feature. If there are |
| 1165 | # more than ten categories, we return top ten (by count) and return |
| 1166 | # one more CategoryCount with category "_OTHER_" and count as |
| 1167 | # aggregate counts of remaining categories. |
| 1168 | { # Represents the count of a single category within the cluster. |
| 1169 | "category": "A String", # The name of category. |
| 1170 | "count": "A String", # The count of training samples matching the category within the |
| 1171 | # cluster. |
| 1172 | }, |
| 1173 | ], |
| 1174 | }, |
| 1175 | }, |
| 1176 | ], |
| 1177 | "centroidId": "A String", # Centroid id. |
| 1178 | }, |
| 1179 | ], |
| 1180 | }, |
| 1181 | }, |
| 1182 | "trainingOptions": { # Options that were used for this training run, includes |
| 1183 | # user specified and default options that were used. |
| 1184 | "dropout": 3.14, # Dropout probability for dnn models. |
| 1185 | "learnRate": 3.14, # Learning rate in training. Used only for iterative training algorithms. |
| 1186 | "labelClassWeights": { # Weights associated with each label class, for rebalancing the |
| 1187 | # training data. Only applicable for classification models. |
| 1188 | "a_key": 3.14, |
| 1189 | }, |
| 1190 | "subsample": 3.14, # Subsample fraction of the training data to grow tree to prevent |
| 1191 | # overfitting for boosted tree models. |
| 1192 | "earlyStop": True or False, # Whether to stop early when the loss doesn't improve significantly |
| 1193 | # any more (compared to min_relative_progress). Used only for iterative |
| 1194 | # training algorithms. |
| 1195 | "dataSplitEvalFraction": 3.14, # The fraction of evaluation data over the whole input data. The rest |
| 1196 | # of data will be used as training data. The format should be double. |
| 1197 | # Accurate to two decimal places. |
| 1198 | # Default value is 0.2. |
| 1199 | "initialLearnRate": 3.14, # Specifies the initial learning rate for the line search learn rate |
| 1200 | # strategy. |
| 1201 | "itemColumn": "A String", # Item column specified for matrix factorization models. |
| 1202 | "inputLabelColumns": [ # Name of input label columns in training data. |
| 1203 | "A String", |
| 1204 | ], |
| 1205 | "warmStart": True or False, # Whether to train a model from the last checkpoint. |
| 1206 | "learnRateStrategy": "A String", # The strategy to determine learn rate for the current iteration. |
| 1207 | "numFactors": "A String", # Num factors specified for matrix factorization models. |
| 1208 | "lossType": "A String", # Type of loss function used during training run. |
| 1209 | "hiddenUnits": [ # Hidden units for dnn models. |
| 1210 | "A String", |
| 1211 | ], |
| 1212 | "kmeansInitializationMethod": "A String", # The method used to initialize the centroids for kmeans algorithm. |
| 1213 | "l1Regularization": 3.14, # L1 regularization coefficient. |
| 1214 | "distanceType": "A String", # Distance type for clustering models. |
| 1215 | "walsAlpha": 3.14, # Hyperparameter for matrix factoration when implicit feedback type is |
| 1216 | # specified. |
| 1217 | "feedbackType": "A String", # Feedback type that specifies which algorithm to run for matrix |
| 1218 | # factorization. |
| 1219 | "optimizationStrategy": "A String", # Optimization strategy for training linear regression models. |
| 1220 | "dataSplitColumn": "A String", # The column to split data with. This column won't be used as a |
| 1221 | # feature. |
| 1222 | # 1. When data_split_method is CUSTOM, the corresponding column should |
| 1223 | # be boolean. The rows with true value tag are eval data, and the false |
| 1224 | # are training data. |
| 1225 | # 2. When data_split_method is SEQ, the first DATA_SPLIT_EVAL_FRACTION |
| 1226 | # rows (from smallest to largest) in the corresponding column are used |
| 1227 | # as training data, and the rest are eval data. It respects the order |
| 1228 | # in Orderable data types: |
| 1229 | # https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#data-type-properties |
| 1230 | "maxIterations": "A String", # The maximum number of iterations in training. Used only for iterative |
| 1231 | # training algorithms. |
| 1232 | "userColumn": "A String", # User column specified for matrix factorization models. |
| 1233 | "maxTreeDepth": "A String", # Maximum depth of a tree for boosted tree models. |
| 1234 | "l2Regularization": 3.14, # L2 regularization coefficient. |
| 1235 | "modelUri": "A String", # [Beta] Google Cloud Storage URI from which the model was imported. Only |
| 1236 | # applicable for imported models. |
| 1237 | "batchSize": "A String", # Batch size for dnn models. |
| 1238 | "minRelativeProgress": 3.14, # When early_stop is true, stops training when accuracy improvement is |
| 1239 | # less than 'min_relative_progress'. Used only for iterative training |
| 1240 | # algorithms. |
| 1241 | "kmeansInitializationColumn": "A String", # The column used to provide the initial centroids for kmeans algorithm |
| 1242 | # when kmeans_initialization_method is CUSTOM. |
| 1243 | "numClusters": "A String", # Number of clusters for clustering models. |
| 1244 | "dataSplitMethod": "A String", # The data split type for training and evaluation, e.g. RANDOM. |
| 1245 | "minSplitLoss": 3.14, # Minimum split loss for boosted tree models. |
| 1246 | }, |
| 1247 | "dataSplitResult": { # Data split result. This contains references to the training and evaluation # Data split result of the training run. Only set when the input data is |
| 1248 | # actually split. |
| 1249 | # data tables that were used to train the model. |
| 1250 | "trainingTable": { # Table reference of the training data after split. |
| 1251 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 1252 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 1253 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| 1254 | }, |
| 1255 | "evaluationTable": { # Table reference of the evaluation data after split. |
| 1256 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 1257 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 1258 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| 1259 | }, |
| 1260 | }, |
| 1261 | }, |
| 1262 | ], |
| 1263 | "modelReference": { # Required. Unique identifier for this model. |
| 1264 | "datasetId": "A String", # [Required] The ID of the dataset containing this model. |
| 1265 | "modelId": "A String", # [Required] The ID of the model. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 1266 | "projectId": "A String", # [Required] The ID of the project containing this model. |
| 1267 | }, |
| 1268 | "description": "A String", # Optional. A user-friendly description of this model. |
| 1269 | "etag": "A String", # Output only. A hash of this resource. |
| 1270 | "creationTime": "A String", # Output only. The time when this model was created, in millisecs since the epoch. |
| 1271 | "encryptionConfiguration": { # Custom encryption configuration (e.g., Cloud KMS keys). This shows the |
| 1272 | # encryption configuration of the model data while stored in BigQuery |
| 1273 | # storage. This field can be used with PatchModel to update encryption key |
| 1274 | # for an already encrypted model. |
| 1275 | "kmsKeyName": "A String", # [Optional] Describes the Cloud KMS encryption key that will be used to protect destination BigQuery table. The BigQuery Service Account associated with your project requires access to this encryption key. |
| 1276 | }, |
| 1277 | } |
| 1278 | |
| 1279 | |
| 1280 | Returns: |
| 1281 | An object of the form: |
| 1282 | |
| 1283 | { |
| 1284 | "location": "A String", # Output only. The geographic location where the model resides. This value |
| 1285 | # is inherited from the dataset. |
| 1286 | "friendlyName": "A String", # Optional. A descriptive name for this model. |
| 1287 | "lastModifiedTime": "A String", # Output only. The time when this model was last modified, in millisecs since the epoch. |
| 1288 | "labels": { # The labels associated with this model. You can use these to organize |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1289 | # and group your models. Label keys and values can be no longer |
| 1290 | # than 63 characters, can only contain lowercase letters, numeric |
| 1291 | # characters, underscores and dashes. International characters are allowed. |
| 1292 | # Label values are optional. Label keys must start with a letter and each |
| 1293 | # label in the list must have a different key. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1294 | "a_key": "A String", |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1295 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1296 | "labelColumns": [ # Output only. Label columns that were used to train this model. |
| 1297 | # The output of the model will have a "predicted_" prefix to these columns. |
| 1298 | { # A field or a column. |
| 1299 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 1300 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 1301 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 1302 | # in this case the output parameter does not have this "type" field). |
| 1303 | # Examples: |
| 1304 | # INT64: {type_kind="INT64"} |
| 1305 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 1306 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 1307 | # {type_kind="STRUCT", |
| 1308 | # struct_type={fields=[ |
| 1309 | # {name="x", type={type_kind="STRING"}}, |
| 1310 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 1311 | # ]}} |
| 1312 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 1313 | "fields": [ |
| 1314 | # Object with schema name: StandardSqlField |
| 1315 | ], |
| 1316 | }, |
| 1317 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 1318 | "typeKind": "A String", # Required. The top level type of this field. |
| 1319 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 1320 | }, |
| 1321 | }, |
| 1322 | ], |
| 1323 | "modelType": "A String", # Output only. Type of the model resource. |
| 1324 | "featureColumns": [ # Output only. Input feature columns that were used to train this model. |
| 1325 | { # A field or a column. |
| 1326 | "name": "A String", # Optional. The name of this field. Can be absent for struct fields. |
| 1327 | "type": { # The type of a variable, e.g., a function argument. # Optional. The type of this parameter. Absent if not explicitly |
| 1328 | # specified (e.g., CREATE FUNCTION statement can omit the return type; |
| 1329 | # in this case the output parameter does not have this "type" field). |
| 1330 | # Examples: |
| 1331 | # INT64: {type_kind="INT64"} |
| 1332 | # ARRAY<STRING>: {type_kind="ARRAY", array_element_type="STRING"} |
| 1333 | # STRUCT<x STRING, y ARRAY<DATE>>: |
| 1334 | # {type_kind="STRUCT", |
| 1335 | # struct_type={fields=[ |
| 1336 | # {name="x", type={type_kind="STRING"}}, |
| 1337 | # {name="y", type={type_kind="ARRAY", array_element_type="DATE"}} |
| 1338 | # ]}} |
| 1339 | "structType": { # The fields of this struct, in order, if type_kind = "STRUCT". |
| 1340 | "fields": [ |
| 1341 | # Object with schema name: StandardSqlField |
| 1342 | ], |
| 1343 | }, |
| 1344 | "arrayElementType": # Object with schema name: StandardSqlDataType # The type of the array's elements, if type_kind = "ARRAY". |
| 1345 | "typeKind": "A String", # Required. The top level type of this field. |
| 1346 | # Can be any standard SQL data type (e.g., "INT64", "DATE", "ARRAY"). |
| 1347 | }, |
| 1348 | }, |
| 1349 | ], |
| 1350 | "expirationTime": "A String", # Optional. The time when this model expires, in milliseconds since the epoch. |
| 1351 | # If not present, the model will persist indefinitely. Expired models |
| 1352 | # will be deleted and their storage reclaimed. The defaultTableExpirationMs |
| 1353 | # property of the encapsulating dataset can be used to set a default |
| 1354 | # expirationTime on newly created models. |
| 1355 | "trainingRuns": [ # Output only. Information for all training runs in increasing order of start_time. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1356 | { # Information about a single training query run for the model. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1357 | "startTime": "A String", # The start time of this training run. |
| 1358 | "results": [ # Output of each iteration run, results.size() <= max_iterations. |
| 1359 | { # Information about a single iteration of the training run. |
| 1360 | "trainingLoss": 3.14, # Loss computed on the training data at the end of iteration. |
| 1361 | "evalLoss": 3.14, # Loss computed on the eval data at the end of iteration. |
| 1362 | "index": 42, # Index of the iteration, 0 based. |
| 1363 | "learnRate": 3.14, # Learn rate used for this iteration. |
| 1364 | "durationMs": "A String", # Time taken to run the iteration in milliseconds. |
| 1365 | "arimaResult": { # (Auto-)arima fitting result. Wrap everything in ArimaResult for easier |
| 1366 | # refactoring if we want to use model-specific iteration results. |
| 1367 | "arimaModelInfo": [ # This message is repeated because there are multiple arima models |
| 1368 | # fitted in auto-arima. For non-auto-arima model, its size is one. |
| 1369 | { # Arima model information. |
| 1370 | "arimaFittingMetrics": { # ARIMA model fitting metrics. # Arima fitting metrics. |
| 1371 | "aic": 3.14, # AIC. |
| 1372 | "logLikelihood": 3.14, # Log-likelihood. |
| 1373 | "variance": 3.14, # Variance. |
| 1374 | }, |
| 1375 | "timeSeriesId": "A String", # The id to indicate different time series. |
| 1376 | "arimaCoefficients": { # Arima coefficients. # Arima coefficients. |
| 1377 | "movingAverageCoefficients": [ # Moving-average coefficients, an array of double. |
| 1378 | 3.14, |
| 1379 | ], |
| 1380 | "autoRegressiveCoefficients": [ # Auto-regressive coefficients, an array of double. |
| 1381 | 3.14, |
| 1382 | ], |
| 1383 | "interceptCoefficient": 3.14, # Intercept coefficient, just a double not an array. |
| 1384 | }, |
| 1385 | "hasDrift": True or False, # Whether Arima model fitted with drift or not. It is always false |
| 1386 | # when d is not 1. |
| 1387 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported |
| 1388 | # for one time series. |
| 1389 | "A String", |
| 1390 | ], |
| 1391 | "nonSeasonalOrder": { # Arima order, can be used for both non-seasonal and seasonal parts. # Non-seasonal order. |
| 1392 | "q": "A String", # Order of the moving-average part. |
| 1393 | "d": "A String", # Order of the differencing part. |
| 1394 | "p": "A String", # Order of the autoregressive part. |
| 1395 | }, |
| 1396 | }, |
| 1397 | ], |
| 1398 | "seasonalPeriods": [ # Seasonal periods. Repeated because multiple periods are supported for |
| 1399 | # one time series. |
| 1400 | "A String", |
| 1401 | ], |
| 1402 | }, |
| 1403 | "clusterInfos": [ # Information about top clusters for clustering models. |
| 1404 | { # Information about a single cluster for clustering model. |
| 1405 | "clusterSize": "A String", # Cluster size, the total number of points assigned to the cluster. |
| 1406 | "centroidId": "A String", # Centroid id. |
| 1407 | "clusterRadius": 3.14, # Cluster radius, the average distance from centroid |
| 1408 | # to each point assigned to the cluster. |
| 1409 | }, |
| 1410 | ], |
| 1411 | }, |
| 1412 | ], |
| 1413 | "evaluationMetrics": { # Evaluation metrics of a model. These are either computed on all training # The evaluation metrics over training/eval data that were computed at the |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1414 | # end of training. |
| 1415 | # data or just the eval data based on whether eval data was used during |
| 1416 | # training. These are not present for imported models. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1417 | "binaryClassificationMetrics": { # Evaluation metrics for binary classification/classifier models. # Populated for binary classification/classifier models. |
| 1418 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 1419 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 1420 | # macro-averaged, the metrics are calculated for each label and then an |
| 1421 | # unweighted average is taken of those values. When micro-averaged, the |
| 1422 | # metric is calculated globally by counting the total number of correctly |
| 1423 | # predicted rows. |
| 1424 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 1425 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 1426 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 1427 | # classification models this is the positive class threshold. |
| 1428 | # For multi-class classfication models this is the confidence |
| 1429 | # threshold. |
| 1430 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 1431 | # metric. |
| 1432 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 1433 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 1434 | # this is a macro-averaged metric. |
| 1435 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 1436 | # positive actual labels. For multiclass this is a macro-averaged |
| 1437 | # metric treating each class as a binary classifier. |
| 1438 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 1439 | # multiclass this is a micro-averaged metric. |
| 1440 | }, |
| 1441 | "negativeLabel": "A String", # Label representing the negative class. |
| 1442 | "positiveLabel": "A String", # Label representing the positive class. |
| 1443 | "binaryConfusionMatrixList": [ # Binary confusion matrix at multiple thresholds. |
| 1444 | { # Confusion matrix for binary classification models. |
| 1445 | "falseNegatives": "A String", # Number of false samples predicted as false. |
| 1446 | "falsePositives": "A String", # Number of false samples predicted as true. |
| 1447 | "trueNegatives": "A String", # Number of true samples predicted as false. |
| 1448 | "f1Score": 3.14, # The equally weighted average of recall and precision. |
| 1449 | "precision": 3.14, # The fraction of actual positive predictions that had positive actual |
| 1450 | # labels. |
| 1451 | "positiveClassThreshold": 3.14, # Threshold value used when computing each of the following metric. |
| 1452 | "accuracy": 3.14, # The fraction of predictions given the correct label. |
| 1453 | "truePositives": "A String", # Number of true samples predicted as true. |
| 1454 | "recall": 3.14, # The fraction of actual positive labels that were given a positive |
| 1455 | # prediction. |
| 1456 | }, |
| 1457 | ], |
| 1458 | }, |
| 1459 | "regressionMetrics": { # Evaluation metrics for regression and explicit feedback type matrix # Populated for regression models and explicit feedback type matrix |
| 1460 | # factorization models. |
| 1461 | # factorization models. |
| 1462 | "meanSquaredError": 3.14, # Mean squared error. |
| 1463 | "rSquared": 3.14, # R^2 score. |
| 1464 | "medianAbsoluteError": 3.14, # Median absolute error. |
| 1465 | "meanSquaredLogError": 3.14, # Mean squared log error. |
| 1466 | "meanAbsoluteError": 3.14, # Mean absolute error. |
| 1467 | }, |
| 1468 | "rankingMetrics": { # Evaluation metrics used by weighted-ALS models specified by # [Alpha] Populated for implicit feedback type matrix factorization |
| 1469 | # models. |
| 1470 | # feedback_type=implicit. |
| 1471 | "meanAveragePrecision": 3.14, # Calculates a precision per user for all the items by ranking them and |
| 1472 | # then averages all the precisions across all the users. |
| 1473 | "normalizedDiscountedCumulativeGain": 3.14, # A metric to determine the goodness of a ranking calculated from the |
| 1474 | # predicted confidence by comparing it to an ideal rank measured by the |
| 1475 | # original ratings. |
| 1476 | "averageRank": 3.14, # Determines the goodness of a ranking by computing the percentile rank |
| 1477 | # from the predicted confidence and dividing it by the original rank. |
| 1478 | "meanSquaredError": 3.14, # Similar to the mean squared error computed in regression and explicit |
| 1479 | # recommendation models except instead of computing the rating directly, |
| 1480 | # the output from evaluate is computed against a preference which is 1 or 0 |
| 1481 | # depending on if the rating exists or not. |
| 1482 | }, |
| 1483 | "multiClassClassificationMetrics": { # Evaluation metrics for multi-class classification/classifier models. # Populated for multi-class classification/classifier models. |
| 1484 | "aggregateClassificationMetrics": { # Aggregate metrics for classification/classifier models. For multi-class # Aggregate classification metrics. |
| 1485 | # models, the metrics are either macro-averaged or micro-averaged. When |
| 1486 | # macro-averaged, the metrics are calculated for each label and then an |
| 1487 | # unweighted average is taken of those values. When micro-averaged, the |
| 1488 | # metric is calculated globally by counting the total number of correctly |
| 1489 | # predicted rows. |
| 1490 | "recall": 3.14, # Recall is the fraction of actual positive labels that were given a |
| 1491 | # positive prediction. For multiclass this is a macro-averaged metric. |
| 1492 | "threshold": 3.14, # Threshold at which the metrics are computed. For binary |
| 1493 | # classification models this is the positive class threshold. |
| 1494 | # For multi-class classfication models this is the confidence |
| 1495 | # threshold. |
| 1496 | "rocAuc": 3.14, # Area Under a ROC Curve. For multiclass this is a macro-averaged |
| 1497 | # metric. |
| 1498 | "logLoss": 3.14, # Logarithmic Loss. For multiclass this is a macro-averaged metric. |
| 1499 | "f1Score": 3.14, # The F1 score is an average of recall and precision. For multiclass |
| 1500 | # this is a macro-averaged metric. |
| 1501 | "precision": 3.14, # Precision is the fraction of actual positive predictions that had |
| 1502 | # positive actual labels. For multiclass this is a macro-averaged |
| 1503 | # metric treating each class as a binary classifier. |
| 1504 | "accuracy": 3.14, # Accuracy is the fraction of predictions given the correct label. For |
| 1505 | # multiclass this is a micro-averaged metric. |
| 1506 | }, |
| 1507 | "confusionMatrixList": [ # Confusion matrix at different thresholds. |
| 1508 | { # Confusion matrix for multi-class classification models. |
| 1509 | "confidenceThreshold": 3.14, # Confidence threshold used when computing the entries of the |
| 1510 | # confusion matrix. |
| 1511 | "rows": [ # One row per actual label. |
| 1512 | { # A single row in the confusion matrix. |
| 1513 | "entries": [ # Info describing predicted label distribution. |
| 1514 | { # A single entry in the confusion matrix. |
| 1515 | "predictedLabel": "A String", # The predicted label. For confidence_threshold > 0, we will |
| 1516 | # also add an entry indicating the number of items under the |
| 1517 | # confidence threshold. |
| 1518 | "itemCount": "A String", # Number of items being predicted as this label. |
| 1519 | }, |
| 1520 | ], |
| 1521 | "actualLabel": "A String", # The original label of this row. |
| 1522 | }, |
| 1523 | ], |
| 1524 | }, |
| 1525 | ], |
| 1526 | }, |
| 1527 | "clusteringMetrics": { # Evaluation metrics for clustering models. # Populated for clustering models. |
| 1528 | "meanSquaredDistance": 3.14, # Mean of squared distances between each sample to its cluster centroid. |
| 1529 | "daviesBouldinIndex": 3.14, # Davies-Bouldin index. |
| 1530 | "clusters": [ # [Beta] Information for all clusters. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1531 | { # Message containing the information about one cluster. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1532 | "count": "A String", # Count of training data rows that were assigned to this cluster. |
| 1533 | "featureValues": [ # Values of highly variant features for this cluster. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1534 | { # Representative value of a single feature within the cluster. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1535 | "numericalValue": 3.14, # The numerical feature value. This is the centroid value for this |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1536 | # feature. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1537 | "featureColumn": "A String", # The feature column name. |
| 1538 | "categoricalValue": { # Representative value of a categorical feature. # The categorical feature value. |
| 1539 | "categoryCounts": [ # Counts of all categories for the categorical feature. If there are |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1540 | # more than ten categories, we return top ten (by count) and return |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1541 | # one more CategoryCount with category "_OTHER_" and count as |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1542 | # aggregate counts of remaining categories. |
| 1543 | { # Represents the count of a single category within the cluster. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1544 | "category": "A String", # The name of category. |
| 1545 | "count": "A String", # The count of training samples matching the category within the |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1546 | # cluster. |
| 1547 | }, |
| 1548 | ], |
| 1549 | }, |
| 1550 | }, |
| 1551 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1552 | "centroidId": "A String", # Centroid id. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1553 | }, |
| 1554 | ], |
| 1555 | }, |
| 1556 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1557 | "trainingOptions": { # Options that were used for this training run, includes |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1558 | # user specified and default options that were used. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1559 | "dropout": 3.14, # Dropout probability for dnn models. |
| 1560 | "learnRate": 3.14, # Learning rate in training. Used only for iterative training algorithms. |
| 1561 | "labelClassWeights": { # Weights associated with each label class, for rebalancing the |
| 1562 | # training data. Only applicable for classification models. |
| 1563 | "a_key": 3.14, |
| 1564 | }, |
| 1565 | "subsample": 3.14, # Subsample fraction of the training data to grow tree to prevent |
| 1566 | # overfitting for boosted tree models. |
| 1567 | "earlyStop": True or False, # Whether to stop early when the loss doesn't improve significantly |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1568 | # any more (compared to min_relative_progress). Used only for iterative |
| 1569 | # training algorithms. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1570 | "dataSplitEvalFraction": 3.14, # The fraction of evaluation data over the whole input data. The rest |
| 1571 | # of data will be used as training data. The format should be double. |
| 1572 | # Accurate to two decimal places. |
| 1573 | # Default value is 0.2. |
| 1574 | "initialLearnRate": 3.14, # Specifies the initial learning rate for the line search learn rate |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1575 | # strategy. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1576 | "itemColumn": "A String", # Item column specified for matrix factorization models. |
| 1577 | "inputLabelColumns": [ # Name of input label columns in training data. |
| 1578 | "A String", |
| 1579 | ], |
| 1580 | "warmStart": True or False, # Whether to train a model from the last checkpoint. |
| 1581 | "learnRateStrategy": "A String", # The strategy to determine learn rate for the current iteration. |
| 1582 | "numFactors": "A String", # Num factors specified for matrix factorization models. |
| 1583 | "lossType": "A String", # Type of loss function used during training run. |
| 1584 | "hiddenUnits": [ # Hidden units for dnn models. |
| 1585 | "A String", |
| 1586 | ], |
| 1587 | "kmeansInitializationMethod": "A String", # The method used to initialize the centroids for kmeans algorithm. |
| 1588 | "l1Regularization": 3.14, # L1 regularization coefficient. |
| 1589 | "distanceType": "A String", # Distance type for clustering models. |
| 1590 | "walsAlpha": 3.14, # Hyperparameter for matrix factoration when implicit feedback type is |
| 1591 | # specified. |
| 1592 | "feedbackType": "A String", # Feedback type that specifies which algorithm to run for matrix |
| 1593 | # factorization. |
| 1594 | "optimizationStrategy": "A String", # Optimization strategy for training linear regression models. |
| 1595 | "dataSplitColumn": "A String", # The column to split data with. This column won't be used as a |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1596 | # feature. |
| 1597 | # 1. When data_split_method is CUSTOM, the corresponding column should |
| 1598 | # be boolean. The rows with true value tag are eval data, and the false |
| 1599 | # are training data. |
| 1600 | # 2. When data_split_method is SEQ, the first DATA_SPLIT_EVAL_FRACTION |
| 1601 | # rows (from smallest to largest) in the corresponding column are used |
| 1602 | # as training data, and the rest are eval data. It respects the order |
| 1603 | # in Orderable data types: |
| 1604 | # https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#data-type-properties |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1605 | "maxIterations": "A String", # The maximum number of iterations in training. Used only for iterative |
| 1606 | # training algorithms. |
| 1607 | "userColumn": "A String", # User column specified for matrix factorization models. |
| 1608 | "maxTreeDepth": "A String", # Maximum depth of a tree for boosted tree models. |
| 1609 | "l2Regularization": 3.14, # L2 regularization coefficient. |
| 1610 | "modelUri": "A String", # [Beta] Google Cloud Storage URI from which the model was imported. Only |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1611 | # applicable for imported models. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1612 | "batchSize": "A String", # Batch size for dnn models. |
| 1613 | "minRelativeProgress": 3.14, # When early_stop is true, stops training when accuracy improvement is |
| 1614 | # less than 'min_relative_progress'. Used only for iterative training |
| 1615 | # algorithms. |
| 1616 | "kmeansInitializationColumn": "A String", # The column used to provide the initial centroids for kmeans algorithm |
| 1617 | # when kmeans_initialization_method is CUSTOM. |
| 1618 | "numClusters": "A String", # Number of clusters for clustering models. |
| 1619 | "dataSplitMethod": "A String", # The data split type for training and evaluation, e.g. RANDOM. |
| 1620 | "minSplitLoss": 3.14, # Minimum split loss for boosted tree models. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1621 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1622 | "dataSplitResult": { # Data split result. This contains references to the training and evaluation # Data split result of the training run. Only set when the input data is |
| 1623 | # actually split. |
| 1624 | # data tables that were used to train the model. |
| 1625 | "trainingTable": { # Table reference of the training data after split. |
| 1626 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 1627 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 1628 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1629 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1630 | "evaluationTable": { # Table reference of the evaluation data after split. |
| 1631 | "tableId": "A String", # [Required] The ID of the table. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 1632 | "projectId": "A String", # [Required] The ID of the project containing this table. |
| 1633 | "datasetId": "A String", # [Required] The ID of the dataset containing this table. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1634 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1635 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1636 | }, |
| 1637 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1638 | "modelReference": { # Required. Unique identifier for this model. |
| 1639 | "datasetId": "A String", # [Required] The ID of the dataset containing this model. |
| 1640 | "modelId": "A String", # [Required] The ID of the model. The ID must contain only letters (a-z, A-Z), numbers (0-9), or underscores (_). The maximum length is 1,024 characters. |
| 1641 | "projectId": "A String", # [Required] The ID of the project containing this model. |
| 1642 | }, |
| 1643 | "description": "A String", # Optional. A user-friendly description of this model. |
| 1644 | "etag": "A String", # Output only. A hash of this resource. |
| 1645 | "creationTime": "A String", # Output only. The time when this model was created, in millisecs since the epoch. |
| 1646 | "encryptionConfiguration": { # Custom encryption configuration (e.g., Cloud KMS keys). This shows the |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1647 | # encryption configuration of the model data while stored in BigQuery |
| 1648 | # storage. This field can be used with PatchModel to update encryption key |
| 1649 | # for an already encrypted model. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1650 | "kmsKeyName": "A String", # [Optional] Describes the Cloud KMS encryption key that will be used to protect destination BigQuery table. The BigQuery Service Account associated with your project requires access to this encryption key. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1651 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1652 | }</pre> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1653 | </div> |
| 1654 | |
| 1655 | </body></html> |