Skip to content

MIP-378 Logistic regression

Kostas FILIPPOPOLITIS requested to merge dev/MIP-378/logistic-regression into master

Created by: jassak

  • Logistic regression: implementation and validation tests (closes MIP-558 and MIP-559)
  • Implementation of transformer class LabelBinarizer for transforming a table of non-numeric/non-binary labels into binary, used as a preprocessing step in logistic regression and possibly other classification algorithms
  • Test case generator enhancements:
    • Read data from csv/cde files instead of database (closes MIP-617)
    • Re-implement some randomness generation methods for better results (closes MIP-651)
  • Add support for two enum types in algorithm parameters. This is probably a temporary fix until we have a complete solution. (new ticket MIP-652)
  • Merge Node and Controller exceptions into a single shared module mipengine/exceptions.py
  • Add dev package pytests-subtests
  • Smaller fixes

Merge request reports