Identification of alcohol addicts among high school students using decision tree based algorithm

Abstract

The paper aims to apply a decision tree based machine learning algorithm to predict possible alcohol addicts among high school students. The data mining process is performed on the real-world data collected in two high schools in Portugal. The dataset is originally designed for the estimation of high school student’s performance where alcohol consumption is used as one of the parameters. In the implementation phase, KNIME analytics platform is applied to test the model. The significant part represents preprocessing of data where the new attributes are derived including class attribute labeled using alcohol addict matrix. Afterwards, the linear correlation is used to reduce the number of features. Data processing consists of dividing the dataset into training and test data, making artificial data for training phase and lastly analyzing the outputs of decision tree learner and predictor. Constructed decision tree determines the connections between certain attributes and student alcohol consumption. Finally, the overall accuracy of the model is measured using a confusion matrix.