Classification of alcohol consumption among secondary school students

Abstract

In 2016, the National Institute of Health reported that 26% of 8th graders, 47% of 10th graders, and 64% of 12th graders have all had experience in consuming alcoholic drinks. This finding indicates an accelerating trend in alcohol use among school students, hence a growing concerns among the public. To address this issue, this paper is set to model the alcohol consumption data among the secondary school students and attempt to predict the alcohol consumption behaviors among them. A set of classification experiments are carried out and the classification accuracies are compared between two variations of neural network algorithms; a self-tuning multilayer perceptron classifier (AutoMLP) against the standard MLP using the student alcohol consumption dataset. It is found that AutoMLP produced better accuracy of 64.54% than neural network with 61.78%.