Machine Learning Model for the Prediction of Emotions in a Mobile Application

Abstract

Emotional intelligence is a transversal axis in the integral development of the person. As a variable, in the interaction environment, research is carried out in various fields of science, reaching great advances with the contribution of artificial intelligence. The EMODIANA has been designed as a subjective emotion research tool, with an extension in the EmoAppPro mobile app. The objective is to design and put into production a machine learning model for classification and prediction from the target of emotions, to a simplification (positive, neutral and negative) in real time. The experimentation is carried out with biology students (n = 30) of the Technical Private University of Loja, during an evaluation process, capturing all the video interaction. A thesaurus of emotions is constructed, from observation of experts, with time windows (t = 20 s), applying Fleap's Kappa for label validation. The Knowledge Discovery in Databases -KDD methodology is applied in the ML process, obtaining better results with a model based on a decision tree.