v0.9.0

[pzivich]

v0.9.0

MAJOR UPDATE:

• Python 3.5 will no longer be supported. Version 3.6+ will be added

Functionality Updates:

• Adding Support for DAG Causal diagram support #123 Adding Directed Acyclic Graph Support #141

• Adding cross-fit estimators Add cross-fitting estimators #116

• Single/Double Crossfit AIPTW

• Single/Double Crossfit TMLE

• EmpricialMeanSL estimator consistent with sklearn and supylearner

• SuperLearner addition

• Adding Zipper Plots for simulation studies

• Add custom_models to AIPTW

Housekeeping

• Creating function to load the data and check missingness

• Making printed results look nicer

• Using np.clip use other bounding strategy Replace with np.clip #138

• Warning for IPTW variance for ATT and ATU IPTW Variance #135

• Replacing print_results with verbose Replace print-results with verbose #129

• Return IDs for MonteCarloGFormula MonteCarloGFormula return user-specified IDs #126

Documentation Updates

• Cross-fit estimators

• Zipper plots

[pzivich]

I should break out the AIPW estimation equation into its own function to be within zepid.causal.utils.py. For cross-fitting ease, it should also have the optional argument splits. When None it gives the overall variance, otherwise it will use the variance in each split