An Approximation to Deep Learning Touristic-Related Time Series Forecasting

Abstract

Tourism is one of the biggest economic activities around the world. This means that an adequate planning of existing resources becomes crucial. Precise demand-related forecasting greatly improves this planning. Deep Learning models are showing an greatly improvement on time-series forecasting, particularly the LSTM, which is designed for this kind of tasks. This article introduces the touristic time-series forecasting using LSTM, and compares its accuracy against well known models RandomForest and ARIMA. Our results shows that new LSTM models achieve the best accuracy.