
Categorical variables with high cardinality and sparsity pose challenges for causal inference. The authors propose a method called CAVIAR that embeds categorical variables into a lower-dimensional space to enable stable and robust estimates through dimensionality reduction.


coremsg

embedding-categorical-variables-to-improve-causal-inference-in-high-dimensional-and-sparse-data


Embedding Categorical Variables to Improve Causal Inference in High-Dimensional and Sparse Data
