Aplicación de un modelo de minería de datos para predecir el posible consumo de drogas

Abstract

Abstract: This degree work is focused on the application of data mining techniques to predict possible drugs consumption, it used on information collected by the CEPRA XII project: Automatic processing techniques applied to the analysis and prediction of tobacco consumption. drugs , which contains data related to the consumption of tobacco and other drugs, this information is organized into health variables, sociodemographic variables, tests and standardized questionnaires, a data treatment was carried out applying different types of normalization, for the creation of the models, the supervised learning algorithms were selected: Decision_tree, GaussianNB and Logistic_regression, experiments were carried out with the data submitted to different types of normalization, the metrics were applied: Accuracy, Recall, F1-score and mean square error to evaluate the results of each model. Finally, a web prototype was developed in order to load the datasets and visualize the results of each model.It is shown that it is feasible to apply data mining techniques to predict tobacco consumption on the information provided bythe CEPRA project, since the Logistic_regresion and Decision_tree algorithm obtained quite significant results, likewise it is verified that the results of each model vary according to the type of normalization to which the data were subjected.