Welcome to pycausal explorer’s documentation!
Hello, and welcome to our docs! pycausal explorer is a causal inference library mostly focused on implementing models that predict causal effect. If you don’t know what causal inference is, you can check the definition or study it more comprehensively using brandy neal’s materials.
Below we outline the libraries models and provide a Model Guide on which to pick. If you just want to see how it works, check out our Example.
Install
To install pycausal explorer, simply run
$ pip install pycausal_explorer
Models
Most of the features this library offers are models that predict causal effect. If you want to decide which model to use, check out out Model Guide. If you want to know every model we have, we also have a Model List.
Linear Learners
Linear learners are simple models, that work best on linearly generated data.
Meta learners
Meta learners make use of other machine learning models to predict causal effect. Their effectiveness depends on how well the provided model can predict the relevant variables.
Causal Forests
The Causal Forests model uses random forests as well as K Nearest Neighbors to estimate treatment effect on complex, non-linear situations.
Nearest Neighbors
The Nearest Neighbors model will find the most similar element in the control and treatment groups, and use their difference to find out the effect of treatment.
Propensity score
The propensity score model aims to represent all covariates as a single scalar, the propensity for treatment to occur. This simplifies the estimation of treatment effect.
Datasets
Pycausal explorer also offers datasets to validate causal inference models. Check them out here
Indices
Models: