Similarity encoding for learning with dirty categorical variables

Patricio Cerda, Gaël Varoquaux, Balázs Kégl. 2018.

Accepted for publication at Machine Learning journal. Springer

https://hal.inria.fr/hal-01806175

To be presented at ECML PKDD 2018 http://www.ecmlpkdd2018.org

Required packages

• numpy
• scipy
• pandas
• scikit-learn
• dirty_cat (Implementation of Similarity encoder: https://dirty-cat.github.io)

Citation

Patricio Cerda, Gaël Varoquaux, Balázs Kégl (2018). Similarity encoding for learning with dirty categorical variables. Manuscript accepted for publication at: Machine Learning Journal. Springer