Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commitfe7a902

Browse files
feat: Update proto definitions for bigquery/v2 to support new proto fields for BQML. (#817)
PiperOrigin-RevId: 387137741Source-Link:googleapis/googleapis@8962c92Source-Link:googleapis/googleapis-gen@102f1b4
1 parent3c1be14 commitfe7a902

File tree

2 files changed

+107
-9
lines changed

2 files changed

+107
-9
lines changed

‎google/cloud/bigquery_v2/types/model.py‎

Lines changed: 95 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -96,6 +96,8 @@ class Model(proto.Message):
9696
Output only. Label columns that were used to train this
9797
model. The output of the model will have a `predicted_`
9898
prefix to these columns.
99+
best_trial_id (int):
100+
The best trial_id across all training runs.
99101
"""
100102

101103
classModelType(proto.Enum):
@@ -113,6 +115,7 @@ class ModelType(proto.Enum):
113115
ARIMA=11
114116
AUTOML_REGRESSOR=12
115117
AUTOML_CLASSIFIER=13
118+
ARIMA_PLUS=19
116119

117120
classLossType(proto.Enum):
118121
r"""Loss metric to evaluate model training performance."""
@@ -151,6 +154,7 @@ class DataFrequency(proto.Enum):
151154
WEEKLY=5
152155
DAILY=6
153156
HOURLY=7
157+
PER_MINUTE=8
154158

155159
classHolidayRegion(proto.Enum):
156160
r"""Type of supported holiday regions for time series forecasting
@@ -285,7 +289,7 @@ class RegressionMetrics(proto.Message):
285289
median_absolute_error (google.protobuf.wrappers_pb2.DoubleValue):
286290
Median absolute error.
287291
r_squared (google.protobuf.wrappers_pb2.DoubleValue):
288-
R^2 score.
292+
R^2 score. This corresponds to r2_score in ML.EVALUATE.
289293
"""
290294

291295
mean_absolute_error=proto.Field(
@@ -528,7 +532,7 @@ class ClusteringMetrics(proto.Message):
528532
Mean of squared distances between each sample
529533
to its cluster centroid.
530534
clusters (Sequence[google.cloud.bigquery_v2.types.Model.ClusteringMetrics.Cluster]):
531-
[Beta]Information for all clusters.
535+
Information for all clusters.
532536
"""
533537

534538
classCluster(proto.Message):
@@ -697,10 +701,29 @@ class ArimaSingleModelForecastingMetrics(proto.Message):
697701
Is arima model fitted with drift or not. It
698702
is always false when d is not 1.
699703
time_series_id (str):
700-
The id to indicate different time series.
704+
The time_series_id value for this time series. It will be
705+
one of the unique values from the time_series_id_column
706+
specified during ARIMA model training. Only present when
707+
time_series_id_column training option was used.
708+
time_series_ids (Sequence[str]):
709+
The tuple of time_series_ids identifying this time series.
710+
It will be one of the unique tuples of values present in the
711+
time_series_id_columns specified during ARIMA model
712+
training. Only present when time_series_id_columns training
713+
option was used and the order of values here are same as the
714+
order of time_series_id_columns.
701715
seasonal_periods (Sequence[google.cloud.bigquery_v2.types.Model.SeasonalPeriod.SeasonalPeriodType]):
702716
Seasonal periods. Repeated because multiple
703717
periods are supported for one time series.
718+
has_holiday_effect (google.protobuf.wrappers_pb2.BoolValue):
719+
If true, holiday_effect is a part of time series
720+
decomposition result.
721+
has_spikes_and_dips (google.protobuf.wrappers_pb2.BoolValue):
722+
If true, spikes_and_dips is a part of time series
723+
decomposition result.
724+
has_step_changes (google.protobuf.wrappers_pb2.BoolValue):
725+
If true, step_changes is a part of time series decomposition
726+
result.
704727
"""
705728

706729
non_seasonal_order=proto.Field(
@@ -711,9 +734,19 @@ class ArimaSingleModelForecastingMetrics(proto.Message):
711734
)
712735
has_drift=proto.Field(proto.BOOL,number=3,)
713736
time_series_id=proto.Field(proto.STRING,number=4,)
737+
time_series_ids=proto.RepeatedField(proto.STRING,number=9,)
714738
seasonal_periods=proto.RepeatedField(
715739
proto.ENUM,number=5,enum="Model.SeasonalPeriod.SeasonalPeriodType",
716740
)
741+
has_holiday_effect=proto.Field(
742+
proto.MESSAGE,number=6,message=wrappers_pb2.BoolValue,
743+
)
744+
has_spikes_and_dips=proto.Field(
745+
proto.MESSAGE,number=7,message=wrappers_pb2.BoolValue,
746+
)
747+
has_step_changes=proto.Field(
748+
proto.MESSAGE,number=8,message=wrappers_pb2.BoolValue,
749+
)
717750

718751
non_seasonal_order=proto.RepeatedField(
719752
proto.MESSAGE,number=1,message="Model.ArimaOrder",
@@ -901,7 +934,7 @@ class TrainingRun(proto.Message):
901934
"""
902935

903936
classTrainingOptions(proto.Message):
904-
r"""
937+
r"""Options used in model training.
905938
Attributes:
906939
max_iterations (int):
907940
The maximum number of iterations in training.
@@ -972,8 +1005,9 @@ class TrainingOptions(proto.Message):
9721005
num_clusters (int):
9731006
Number of clusters for clustering models.
9741007
model_uri (str):
975-
[Beta] Google Cloud Storage URI from which the model was
976-
imported. Only applicable for imported models.
1008+
Google Cloud Storage URI from which the model
1009+
was imported. Only applicable for imported
1010+
models.
9771011
optimization_strategy (google.cloud.bigquery_v2.types.Model.OptimizationStrategy):
9781012
Optimization strategy for training linear
9791013
regression models.
@@ -1030,8 +1064,11 @@ class TrainingOptions(proto.Message):
10301064
If a valid value is specified, then holiday
10311065
effects modeling is enabled.
10321066
time_series_id_column (str):
1033-
The id column that will be used to indicate
1034-
different time series to forecast in parallel.
1067+
The time series id column that was used
1068+
during ARIMA model training.
1069+
time_series_id_columns (Sequence[str]):
1070+
The time series id columns that were used
1071+
during ARIMA model training.
10351072
horizon (int):
10361073
The number of periods ahead that need to be
10371074
forecasted.
@@ -1042,6 +1079,15 @@ class TrainingOptions(proto.Message):
10421079
output feature name is A.b.
10431080
auto_arima_max_order (int):
10441081
The max value of non-seasonal p and q.
1082+
decompose_time_series (google.protobuf.wrappers_pb2.BoolValue):
1083+
If true, perform decompose time series and
1084+
save the results.
1085+
clean_spikes_and_dips (google.protobuf.wrappers_pb2.BoolValue):
1086+
If true, clean spikes and dips in the input
1087+
time series.
1088+
adjust_step_changes (google.protobuf.wrappers_pb2.BoolValue):
1089+
If true, detect step changes and make data
1090+
adjustment in the input time series.
10451091
"""
10461092

10471093
max_iterations=proto.Field(proto.INT64,number=1,)
@@ -1120,9 +1166,19 @@ class TrainingOptions(proto.Message):
11201166
proto.ENUM,number=42,enum="Model.HolidayRegion",
11211167
)
11221168
time_series_id_column=proto.Field(proto.STRING,number=43,)
1169+
time_series_id_columns=proto.RepeatedField(proto.STRING,number=51,)
11231170
horizon=proto.Field(proto.INT64,number=44,)
11241171
preserve_input_structs=proto.Field(proto.BOOL,number=45,)
11251172
auto_arima_max_order=proto.Field(proto.INT64,number=46,)
1173+
decompose_time_series=proto.Field(
1174+
proto.MESSAGE,number=50,message=wrappers_pb2.BoolValue,
1175+
)
1176+
clean_spikes_and_dips=proto.Field(
1177+
proto.MESSAGE,number=52,message=wrappers_pb2.BoolValue,
1178+
)
1179+
adjust_step_changes=proto.Field(
1180+
proto.MESSAGE,number=53,message=wrappers_pb2.BoolValue,
1181+
)
11261182

11271183
classIterationResult(proto.Message):
11281184
r"""Information about a single iteration of the training run.
@@ -1218,10 +1274,29 @@ class ArimaModelInfo(proto.Message):
12181274
Whether Arima model fitted with drift or not.
12191275
It is always false when d is not 1.
12201276
time_series_id (str):
1221-
The id to indicate different time series.
1277+
The time_series_id value for this time series. It will be
1278+
one of the unique values from the time_series_id_column
1279+
specified during ARIMA model training. Only present when
1280+
time_series_id_column training option was used.
1281+
time_series_ids (Sequence[str]):
1282+
The tuple of time_series_ids identifying this time series.
1283+
It will be one of the unique tuples of values present in the
1284+
time_series_id_columns specified during ARIMA model
1285+
training. Only present when time_series_id_columns training
1286+
option was used and the order of values here are same as the
1287+
order of time_series_id_columns.
12221288
seasonal_periods (Sequence[google.cloud.bigquery_v2.types.Model.SeasonalPeriod.SeasonalPeriodType]):
12231289
Seasonal periods. Repeated because multiple
12241290
periods are supported for one time series.
1291+
has_holiday_effect (google.protobuf.wrappers_pb2.BoolValue):
1292+
If true, holiday_effect is a part of time series
1293+
decomposition result.
1294+
has_spikes_and_dips (google.protobuf.wrappers_pb2.BoolValue):
1295+
If true, spikes_and_dips is a part of time series
1296+
decomposition result.
1297+
has_step_changes (google.protobuf.wrappers_pb2.BoolValue):
1298+
If true, step_changes is a part of time series decomposition
1299+
result.
12251300
"""
12261301

12271302
non_seasonal_order=proto.Field(
@@ -1237,11 +1312,21 @@ class ArimaModelInfo(proto.Message):
12371312
)
12381313
has_drift=proto.Field(proto.BOOL,number=4,)
12391314
time_series_id=proto.Field(proto.STRING,number=5,)
1315+
time_series_ids=proto.RepeatedField(proto.STRING,number=10,)
12401316
seasonal_periods=proto.RepeatedField(
12411317
proto.ENUM,
12421318
number=6,
12431319
enum="Model.SeasonalPeriod.SeasonalPeriodType",
12441320
)
1321+
has_holiday_effect=proto.Field(
1322+
proto.MESSAGE,number=7,message=wrappers_pb2.BoolValue,
1323+
)
1324+
has_spikes_and_dips=proto.Field(
1325+
proto.MESSAGE,number=8,message=wrappers_pb2.BoolValue,
1326+
)
1327+
has_step_changes=proto.Field(
1328+
proto.MESSAGE,number=9,message=wrappers_pb2.BoolValue,
1329+
)
12451330

12461331
arima_model_info=proto.RepeatedField(
12471332
proto.MESSAGE,
@@ -1319,6 +1404,7 @@ class ArimaModelInfo(proto.Message):
13191404
label_columns=proto.RepeatedField(
13201405
proto.MESSAGE,number=11,message=standard_sql.StandardSqlField,
13211406
)
1407+
best_trial_id=proto.Field(proto.INT64,number=19,)
13221408

13231409

13241410
classGetModelRequest(proto.Message):

‎google/cloud/bigquery_v2/types/table_reference.py‎

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -36,11 +36,23 @@ class TableReference(proto.Message):
3636
maximum length is 1,024 characters. Certain operations allow
3737
suffixing of the table ID with a partition decorator, such
3838
as ``sample_table$20190123``.
39+
project_id_alternative (Sequence[str]):
40+
The alternative field that will be used when ESF is not able
41+
to translate the received data to the project_id field.
42+
dataset_id_alternative (Sequence[str]):
43+
The alternative field that will be used when ESF is not able
44+
to translate the received data to the project_id field.
45+
table_id_alternative (Sequence[str]):
46+
The alternative field that will be used when ESF is not able
47+
to translate the received data to the project_id field.
3948
"""
4049

4150
project_id=proto.Field(proto.STRING,number=1,)
4251
dataset_id=proto.Field(proto.STRING,number=2,)
4352
table_id=proto.Field(proto.STRING,number=3,)
53+
project_id_alternative=proto.RepeatedField(proto.STRING,number=4,)
54+
dataset_id_alternative=proto.RepeatedField(proto.STRING,number=5,)
55+
table_id_alternative=proto.RepeatedField(proto.STRING,number=6,)
4456

4557

4658
__all__=tuple(sorted(__protobuf__.manifest))

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp