engleski

Peach firmness prediction using optimized regression trees models

The paper focuses on creating accurate model for peach firmness prediction. For this purpose, multiple machine learning models and their optimized variations are developed and compared. Because of its simplicity and robustness, multiple linear regression is used as a base-line model for predicting peach firmness. It assumes a linear relationship between numerical predictors and the outcome. Regression trees is the second developed model. It is a flexible data-driven model that can be used for predicting numerical outcome. The experiment aims to investigate the possibility of improving regression tree model using various metaheuristic optimization techniques implemented in metaheuristicOpt and GA R packages. As a proof of concept, prediction accuracy between multiple linear regression, regression trees and optimized regression trees models is compared. The results show that it is possible to improve the peach firmness prediction accuracy of regression trees model using metaheuristic algorithms.

Peach firmness prediction ; regression trees ; global optimization ; machine learning ; metaheuristics

nije evidentirano