Prediction of Anorexia Risk Based on Decision Tree

Abstract

Anorexia is a kind of eating disorder caused by the cognitive distortions of patients. Under the condition, patients' physical functions can be impaired by extreme weight control behaviors. The prediction of anorexia can help patients with early detection and prediction of the risk of disease, then decrease the damage to the body and the psychological burden. In recent years, machine learning models achieve promising performances in prediction problems. In this paper, after trying on the Adaboost model, Extratree model, and decision tree model, this research decides to use the decision tree model. It is suitable for Boolean-type data classification to do data analysis and hence achieves outstanding performances. By building a decision tree model and referring to the correlation between different features, this research concludes data training. People who have the problem with liver firm and spleen palpable are more likely to gain anorexia. Fatigue and malaise have little relationship with anorexia.