Modelo de predicción de la demanda turística mediante técnicas de machine learning.

Abstract

Abstract: This study aims to predict tourism demand by considering various variables, including temporal, sociodemographic, and behavioral factors of tourists visiting Loja. The goal is to provide individuals and institutions involved in this economic activity with a precise tool to help them define strategies and prepare to receive tourists year-round. Experimentation was conducted using two machine learning algorithms and one deep learning algorithm. The machine learning algorithms proved highly effective in predicting daily tourism demand; however, LSTM did not yield favorable results in the metrics. The CRISP DM methodology was used to develop the model, implementing each of its phases. Finally, the demand predictions were visualized in a local web environment developed using the Flask framework.