Updating to match private code development
Overview
This code update brings the evaluation code up to date with in-house development.
Release Notes
- Data file names now mirror the script names that created the files
- Features on food inspections are now calculated separately
- Features on business inspections are now calculated separately
- The model code merges in the features, does not calculate features
- Added script to adjust the public sanitarian data to match the schema of the private sanitarian file
- More aggressive filtering functions
- Separates out the violation matrix calculation into the parsing step and classification step (which, as it turns out will be useful for the new inspection format)
- Refactoring model result / evaluation steps to accommodate future analysis
Related issues
- adding prefix number to code and data, closes #100
- syncing and updating startup script, closes #101
- split violation matrix calculation into two steps, closes #102
- updated help example to remove unused variable
- adding nokey function, needed for new violation matrix calculation
- guard against too few categories in GenerateOtherLicenseInfo, closes 103
- updating filter functions to match model
- starting work described in #104 to split feature creation
- refactoring code for model compatibility
- simplifying initialization