Research and Application of Genetic Algorithm Based Optimization Neural Network Model for Prediction

In this paper, based on the collected dataset, the outliers are first removed from the data using the Box-Plot method, and then the data are normalized using the Box-Cox transform. Based on the BP neural network, a GA-BP model was built using a genetic algorithm to optimize its weight settings. In this paper, the GA-BP model was used to predict, and r-squared and SMAPE were calculated. the results showed that the r-squared and SMAPE of the training set were 0.8402 and 4.799%, respectively, while the r-squared and SMAPE of the test set were 0.8551 and 3.111%, respectively. The r-squared of the BP neural network for the training and test sets were 0.8317 and 0.8149, respectively. the computational results of the GA-BP algorithm outperformed the BP neural network, indicating that the model was estimated with high accuracy.