A Comparative Study of Machine Learning Models for Predicting Length of Stay in Hospitals
Résumé
There has been a growing interest in recent years in correctly predicting the Length of Stay (LoS) in a hospital setting. Estimating the LoS on patient' admission helps hospitals in planning, controlling costs and, providing better services. In this paper, we consider predicting the LoS as a regression problem for which we implement and compare different Machine Learning (ML) algorithms. Multiple Linear Regression (MLR), Support Vector Machines (SVM), Random Forests (RF), and Gradient Boosting model (GBM) are implemented using an open-source dataset. The methodological process involves a preprocessing step combining data transformation, data standardization, and categorical data encoding. Moreover, the Synthetic Minority Over Sampling Technique for Regression (SMOTER) is applied to handle unbalanced data. Then, ML algorithms are employed, with a hyperpa-rameter tuning phase to obtain optimal coefficients. Finally, Mean Absolute Error (MAE), R-squared (R 2), and Adjusted R-squared (Adjusted R 2) metrics are selected to evaluate the model with parameters.