Predictive models for estimating adolescents with a tendency to alcoholism

Abstract

One of the most consumed drugs worldwide is, without a doubt, alcohol. According to some evidence, young people usually come into contact with alcohol between the ages of 12 and 17, this has led to different investigations in order to understand what patterns can condition alcohol consumption in young people. The objective of this article is to analyze three different pre-dictive models based on Machine Learning, in order to understand which of the analyzed models respond in the best way to the study of the prediction of the tendency to alcoholism in young people. To carry out the analysis, a data set of 521 records has been taken as a base, obtained from Kagle as a model, which was subjected to the analysis of the three models. According to the tests carried out with the predictive models, the Linear Regression model has greater precision with an accuracy of 1.00 compared to 0.95 for the KNN model and 0.98 for the Decision Tree. The study determines in the ROC curves analyzed that the linear regression model achieves better results between the sensitivity of true positives and the specificity of false positives. On the other hand, we must mention that according to the data set analyzed, the predictive indicators are the area where the adolescent lives, the family status in which he grows up, and the availability of free time. Although the study does not claim to be conclusive, it reflects the importance of recognizing protective psychosocial factors in the design and implementation of promotion and prevention programs associated with responsible alcohol consumption and non-violent behavior with adolescents from a salutogenic perspective.