Post-processing
ndj_pipeline.post
Post-model fit reporting of results as plots, summary tables and documentation.
- ndj_pipeline.post.create_continuous_plots(df, reporting_features, model_config)
Create line plot to show how target varies according to feature.
No returns; saves assets to model folder.
- Parameters
df (
DataFrame) – Full, feature rich dataframe, must contain config specified target, and numeric feature columns specified by reporting_featuresreporting_features (
List[str]) – List of features to produce individual plotsmodel_config (
Dict[str,Any]) – Loaded model experiment config
- Return type
None
- ndj_pipeline.post.create_correlation_matrix(df, reporting_features, model_config)
Create correlation matrix between subset of reported features.
No returns; saves assets to model folder.
- Parameters
df (
DataFrame) – Full, feature rich dataframe, must contain config specified target, and numeric feature columns specified by reporting_featuresreporting_features (
List[str]) – List of features to produce individual plotsmodel_config (
Dict[str,Any]) – Loaded model experiment config
- Return type
None
- ndj_pipeline.post.create_metrics_plot(results, model_config, name='')
Produce metrics and scatterplot for results table.
No returns; saves assets to model folder.
- Parameters
results (
DataFrame) – DataFrame with “Actual” and “Prediction” columnsmodel_config (
Dict[str,Any]) – Loaded model experiment configname (
str) – Simple label added to outputs, helpful to distinguish models
- Return type
None
- ndj_pipeline.post.create_univariate_plots(df, reporting_features, model_config)
Create scatterplots with linear fit for each feature against target.
No returns; saves assets to model folder.
- Parameters
df (
DataFrame) – Full, feature rich dataframe, must contain config specified target, and numeric feature columns specified by reporting_featuresreporting_features (
List[str]) – List of features to produce individual plotsmodel_config (
Dict[str,Any]) – Loaded model experiment config
- Return type
None