Expedock AutoML

Expedock's AutoML Library.

Info

This library is still in development. The API is subject to change.

Installation

pip3 install -e .

How to Use

Important

Go to the Sample Usage page for a full example.

First, build the configuration dataclasses for your AutoML pipeline.

from expedock_automl import (
    SnowflakeAccountConfig,
    AWSAccountConfig,
    SnowflakeTableConfig,
    ModelConfig,
    ArtifactsCloudSaveConfig,
)

snowflake_account_config = SnowflakeAccountConfig(...)
aws_account_config = AWSAccountConfig(...)

input_table_config = SnowflakeTableConfig(...)
evaluation_table_config = SnowflakeTableConfig(...)
feature_importance_table_config = SnowflakeTableConfig(...)

model_config = ModelConfig(...)
artifacts_cloud_save_config = ArtifactsCloudSaveConfig(...)

Then, simply run the following to train the AutoML model:

from expedock_automl import automl

automl(
    snowflake_account_config,
    aws_account_config,
    input_table_config,
    evaluation_table_config,
    feature_importance_table_config,
    model_config,
    artifacts_cloud_save_config,
)

After that, you can use the trained model to make predictions on new data:

new_input_table_config = SnowflakeTableConfig(...)
output_table_config = SnowflakeTableConfig(...)

from expedock_automl import predict

predict(
    snowflake_account_config,
    aws_account_config,
    new_input_table_config,
    output_table_config,
    artifacts_cloud_save_config,
)

Results

The predictions, evaluation metrics, and feature importance will be saved to the specified Snowflake tables. And the trained model and other artifacts will be saved to the specified AWS S3 bucket.