Quantile Encoder: Tackling High Cardinality Categorical Features in Regression Problems

Regression problems have been widely studied in machine learning literature resulting a plethora of regression models and performance measures. However, there are few techniques specially dedicated to solve the problem how incorporate categorical features problems. Usually, feature encoders general enough cover both classification This lack specificity results underperforming models. In this paper, we provide an in-depth analysis tackle high cardinality with quantile. Our proposal outperforms state-of-the-art encoders, including traditional statistical mean target encoder, when considering Mean Absolute Error, especially presence long-tailed or skewed distributions. Besides, deal possible overfitting categories small support, our encoder benefits from additive smoothing. Finally, describe expand encoded values by creating set different quantiles. expanded provides more informative output about question, further boosting model.