A Feature Subset Selection Approach For Predicting Smoking Behaviours

Abstract

Identifying smoking behavior holds a significant value for informing patients in the early stages. Due to the complexity of this process, the integration of machine learning can provide healthcare professionals with the necessary support to make accurate predictions regarding smoking behavior. To predict if a person smokes or not, the Lasso feature selection method is implemented to identify and select the most relevant features. Subsequently, a set of final subset features is utilized in conjunction with various machine learning classifiers, including LightGBM, XGBoost, Random Forest, and Multilayer Perceptron to perform the prediction task. This study aims to evaluate different classifiers and identify the one with the best performance. After conducting several tests, based on the results obtained, the Random Forest algorithm has outperformed the others, with an accuracy of 84.73%. Additionally, its training speed is significantly faster than other algorithms.