Chromatic Algorithm Applied To A Non-Linear Regression Model In Time Series Forecasts

Algoritmo Cromático Aplicado A Un Modelo De Regresión No Lineal En Pronósticos De Series De Tiempo

The present work uses the new metaheuristic, a chromatic algorithm for the optimization of time series forecasts through a non-linear regression model. In this new proposal, the chromatic algorithm is handled due to its real coding characteristics and its multiple boot memory, which allow it to be more efficient when minimizing the error in the model's forecasts, to achieve this statistical indicators of the error are used that help improve predictions for each specific problem.

In addition, improvements are devised to the regression model and the algorithm used in such a way that it is possible to predict the behavior of the problems, not only of one variable but also of multiple variables. The algorithm together with the model is tested in different problems with one and multiple variables, providing very good predictions. It is executed in a practical case study related to the estimation of livestock prices according to their type in the studied region. This new method generates more possibilities to achieve that the forecasts are adjusted and to improve any type of prediction. This research provides a new way to minimize forecast errors and generate high-quality results.

It also shows that it is possible to establish forecasts in both single and multi-variable problems, with reasonable computational times. This would be an excellent strategy for the countless companies, entities or organizations that require truly efficient methods that allow them to make the best decisions.