The average prediction accuracies of the stepwise regression model
was apparently less than the latter two models, which indicates that
the result and prediction accuracy of the stepwise regression model
was worse than the latter two models and should be used with caution
in actual cases. The average prediction accuracies of the BP neural
network and SVM were over 93%, which shows they both are reliable
and applied to the actual forecast business. Particularly, the average
prediction accuracies of SVM was larger than that of the BP neural network,
it manifests SVM is better than BP neural network in predicting,
but the difference is not large. The comparison of accuracies of three prediction
models turned out that: SVM N BP neural network N stepwise
regression.