Vizdataquality

Changes made to make reading data from files in chunks and adding nodes to ExplanationGraph much faster, which is important for large datasets (millions of records). Changes also made to allow the data to effectively have an unlimited number of combinations of missing values (tested with a dataset that has 500,000 combinations).

Changed purity calculation to 'overall', corrected purity heatmap rounding issue and a typo.

New functionality and example notebooks for investigating patterns and structures of missing data.

- First full release of vizdataquality, a new package for investigating data quality in a six-step workflow.
- New stackedbar plotting
- Better and more documentation

Initial, alpha release of vizdataquality, a new package for investigating data quality.