Skip to content

Resolve "Create training & dev sets"

Andreas Rubin-Schwarz requested to merge 2-create-training-dev-sets into master

Closes #2 (closed)

  • quick introduction into reasoning behind split
  • discussed ratio choices
  • introduced stratification
  • implemented workaround for continuous variables
  • investigated equal-width and quantile binning
  • implemented Freedman–Diaconis rule and Doane's formula for bin size
  • implemented binning adjustment for bridge-classes
  • stratification analysis and thoughts on complexity
  • adjustments and final splits
Edited by Andreas Rubin-Schwarz

Merge request reports