Workflows

Common patterns for using drx to execute useful tasks in DataRobot.

Spark Ingest

Ingest arbitrarily large Spark dataframes, downsampling if necessary. PROTOTYPE

base = drx.AutoMLModel()
model = drx.SparkIngestModel(base, sampling_strategy='smart')
model.fit(spark_df, target='my_target')

Spark Ingest

Prediction Explanation Clustering

Discover new insights by grouping predictions together based on their explanations

pivot_explanations = drx.featurize_explanations(preds)
model = drx.AutoClusteringModel()
model.fit(pivot_explanations)

Prediction Explanation Clustering

Backtesting (OTV)

Configure time-aware model validation and backtesting for Autopilot

model = drx.AutoMLModel()
drx.Configurator(model).otv(n_backtests=5, v_duration='30d')
model.fit(df, target='my_target')

Backtesting (OTV)

Clustering and Segmented Modeling

Use time series clustering to setup a segmented modeling project

df['Cluster'] = df['seriesID'].map(cluster_mapping)
ts_model = drx.AutoTSModel()
ts_model.fit(df, target="SalesQty", datetime_partition_column="Date", multiseries_id_columns="seriesId", user_defined_segment_id_columns="Cluster")

Time Series Clustering and Segmented Modeling

Integration with MLflow

Use MLflow for experiment tracking and model serving PROTOTYPE

drx.drx_mlflow.autolog()

Integration with MLflow