Analysis utilities for binary classification and Causata users.
The Causata package provides utilities for extracting data from the Causata application, training binary classification models, and exporting models as PMML for scoring.
This file summarizes changes to the Causata package.
- Fixed issue with default protocol for CausataConfig.
- Created distinct username / password arguments for Connect and CausataConfig.
- Dramatically reduced memory usage and increased speed in GetData using data.table
- Added more tests for replaying transformations
- Added more tests for discretization
- Fixed bug with discretization of integer values
- Discretization now requires outlier replacement first
- Added function ShortenStrings for printing variables with long names in BinaryPredictor.
- The predict.GlmnetModelDefinition function now returns a list of missing columns.
- Fixed bug in PMML generation for factor levels with period in the level name.
- Fixed bug in vignette that caused a knitr error.
- Added tests for PMML generation.