Metadata-Version: 2.1
Name: causal-selection
Version: 1.14
Summary: Select and score features for causal inferences
Home-page: https://github.com/rpryzant/causal_selection
Author: Reid Pryzant
Author-email: rpryzant@stanford.edu
License: UNKNOWN
Keywords: feature selection
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.6
Requires-Dist: scipy
Requires-Dist: sklearn
Requires-Dist: nltk
Requires-Dist: pandas
Requires-Dist: torch (>=1.0.0)


This package has two interfaces:

(1) score_vocab(): Given text (T), vocab (V), outcome(s) Y, and 
    confound(s) (C), this method will score each element of the
    vocab according to how well it explains each Y, controlling 
    for all of the C's.

(2) evaluate_vocab(): Measure's the strength of a vocab's causal
    effects on Y (controlling for C).

TODOs
    - loss weighting
    - scheduling
    - layers changeable

(c) Reid Pryzant 2019 https://cs.stanford.edu/~rpryzant/
May be used and distributed under the MIT license.


