rezarokni commented on a change in pull request #13644: URL: https://github.com/apache/beam/pull/13644#discussion_r606037203
########## File path: website/www/site/content/en/documentation/patterns/bqml.md ########## @@ -0,0 +1,98 @@ +--- +title: "BigQuery ML integration" +--- + +<!-- +Licensed under the Apache License, Version 2.0 (the "License"); +you may not use this file except in compliance with the License. +You may obtain a copy of the License at +http://www.apache.org/licenses/LICENSE-2.0 +Unless required by applicable law or agreed to in writing, software +distributed under the License is distributed on an "AS IS" BASIS, +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +See the License for the specific language governing permissions and +limitations under the License. +--> + +# BigQuery ML integration + +With the samples on this page we will demonstrate how to integrate models exported from [BigQuery ML (BQML)](https://cloud.google.com/bigquery-ml/docs) into your Apache Beam pipeline using [TFX Basic Shared Libraries (tfx_bsl)](https://github.com/tensorflow/tfx-bsl). + +Roughly, the sections below will go through the following steps in more detail: + +1. Create and train your BigQuery ML model +1. Export your BigQuery ML model +1. Create a transform that uses the brand-new BigQuery ML model + +## Create and train your BigQuery ML model + +To be able to incorporate your BQML model into an Apache Beam pipeline using tfx_bsl, it has to be in the [TensorFlow SavedModel](https://www.tensorflow.org/guide/saved_model) format. An overview that maps different model types to their export model format for BQML can be found [here](https://cloud.google.com/bigquery-ml/docs/exporting-models#export_model_formats_and_samples). + +For the sake of simplicity, we'll be training a (simplified version of the) logistic regression model in the [BQML quickstart guide](https://cloud.google.com/bigquery-ml/docs/bigqueryml-web-ui-start), using the publicly available Google Analytics sample dataset. An overview of all models you can create using BQML can be found [here](https://cloud.google.com/bigquery-ml/docs/introduction#supported_models_in). + +After creating a BigQuery dataset, you continue to create the model, which is fully defined in SQL: + +```SQL +CREATE MODEL IF NOT EXISTS `bqml_tutorial.sample_model` +OPTIONS(model_type='logistic_reg', input_label_cols=["label"]) AS +SELECT + IF(totals.transactions IS NULL, 0, 1) AS label, + IFNULL(geoNetwork.country, "") AS country +FROM + `bigquery-public-data.google_analytics_sample.ga_sessions_*` +WHERE + _TABLE_SUFFIX BETWEEN '20160801' AND '20170630' Review comment: Can you please make a note on this with a link to the various options that can exist in the BQ table. So that folks know to use the right one dependent on their source. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
