{"repo":"kristeligt-dagblad/dbt_ml","free":true,"listed":false,"github":"https://github.com/kristeligt-dagblad/dbt_ml","clone":"git clone https://github.com/kristeligt-dagblad/dbt_ml.git","description":"Package for dbt that allows users to train, audit and use BigQuery ML models.","language":"Makefile","stars":77,"topics":["dbt","bigquery-ml"],"license":"Apache-2.0","category":"data-pipelines","readme_excerpt":"BigQuery ML models in dbt Package for dbt that allows users to train, audit and use BigQuery ML models. The package implements a model materialization that trains a BigQuery ML model from a select statement and a set of parameters. In addition to the model materialization a set of helper macros that assist with model audit and prediction are included. Installation To install the package add the package path to the packages.yml file in your dbt project In order to use the model audit post-hook the following variables have to be set in your dbt project.yml file. Variable Description --------------------- -------------------------- dbt ml:audit schema Schema of the audit table. dbt ml:audit table Name of the audit table. You will also need to specify the post-hook in your dbt project.yml file [1] as {{ dbt ml.model audit() }} . Optionally, you can create the audit dataset and use the dbt ml.create model audit table() macro to create the audit table automatically if it does not exist. Example config for dbt project.yml below: Usage In order to use the model materialization, simply create a .sql file with a select statement and set the materialization to model . Additionaly, specify any BigQuery ML options in the ml config key of under meta in the model configuration. With dbt-core, ml config can be a top level configuration, not nested under meta . Storing ml config under meta is the recommended new configuration as it is supported both in dbt-core and Fusion. Note that the mater","default_branch":null,"files":null,"tree":[],"storefront":"/r/kristeligt-dagblad","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/kristeligt-dagblad/dbt_ml/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}