Predicción de compras de productos mediante KDD y Aprendizaje Automático.

Abstract

Abstract: In the present work, an investigation is carried out into the prediction of product purchases using Automatic Learning algorithms, which includes Machine Learning and Deep Learning models. The research includes a review of the state of the art, through which the variables, models, metrics and technologies used for purchase prediction are analyzed. Based on this review, the dataset is chosen, which consists of a purchase history of a British company that sells products online; and subsequently, the AI models are trained. The best model obtained is the Support Vector Machine, for a scenario with the seven best variables of the dataset and in which balancing techniques are not applied. Then, the prototype is developed, which includes 5 options: file entry, dataset visualization, dataset transformation, evaluation metrics and results obtained. Finally, the AI model, the prototype and the thesis are validated.