search

sys-bio / SBstoat

Parameter optimization using Tellurium
MIT License
1 stars 4 forks source link

ModelFitter doesn't support cross validation #17

Closed joseph-hellerstein closed 3 years ago

joseph-hellerstein commented 3 years ago

Supporting cross validation means:

  1. Fitting on a subset of the timepoints
  2. Evaluating the fit quality on another subset
joseph-hellerstein commented 3 years ago

Design:

  1. Create CrossValidator that accepts a GenericModel supporting the methods: train, predict, score. The constructor includes number of folds and foldStrategy (e.g., alternating).
  2. ModelFitter does time series alignment by: (a) selecting numPoint; (b) determining the indices of fittedTS that match observedTS
  3. ModelFitter has a crossValidate method GenericModel interface for ModelFitter.
joseph-hellerstein commented 3 years ago

Implemented in release 1.16.