Examinando por Autor "Castro Sanchez, Cristian Jose"
Mostrando 1 - 1 de 1
Resultados por página
Opciones de ordenación
- ÍtemAnálisis y conversión de variable de conteo de uso en variable calificación para la elaboración de un modelo de recomendación de canciones(Fundación Universitaria Los Libertadores. Sede Bogotá., ) Castro Sanchez, Cristian Jose; Gonzalez Veloza., JhonThis article presents a data analysis that allows multiple ways to convert the count variable into a rating variable, in order to generate a song recommendation model based on a classic rock dataset through the collaborative filter of the Turicreate tool. Machine Learning with the help of Ubuntu Python software. As a result, it was found that the database has a high bias, which involved analyzing the data in depth and creating a model that considers with greater weight the data heard more than once by users. The recommendation model has an accuracy of 0.075 and 0.20 recovery higher than the model that precedes it in comparative performance values, this shows that the model provides a satisfactory recommendation whether filtering by user, song title or artist. It was evidenced that the model provides an agile and reliable methodology for the logical understanding of the analysis and creation of a recommendation system. As a general conclusion, it is observed that, among different ways of mitigating the bias.