Can anyone interpret this regression plot for neural networks?

Question

kareema jumoorty 2020-3-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/510075-can-anyone-interpret-this-regression-plot-for-neural-networks

回答： Jesse Nyokabi 2021-2-19

Hello! I am trying a multilayer perceptron for intradaily data. The training is going smoothly but I get the following regression plot. Can anyone tell me if that is acceptable result? Thank you!!

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Nipun Katyal 2020-3-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/510075-can-anyone-interpret-this-regression-plot-for-neural-networks#answer_419962

As the ylabel of these graphs denote an equation between the predicted value and the target value, with output as the dependent variable and target as the independent variable. These equations can be used to show how well your MLP is able to perform. The coefficient of Target shows the proportionality between the output and the targets, hence for a good performance MLP it should be as close to unity as possible. The second term which is a constant, is the error or the residue which should be added to the scaled Target to make it as close as possible to the predicted output, ideally it should be zero or as small as possible. The title represents the coefficient of regression between target and the output. As far as the results for your classifier go, there is some disparity between the training and the testing accuracy, maybe it is because of overfitting, but now you have a clear idea about the plots and can use them to compare the results to find the best results.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Nipun Katyal 2020-3-13

Apart from overfitting hyperparameters the two main causes can be:

The size of the test set is too small.
The test and the train data follow different distributions which explains the difference in the accuracies at the time of validation and testing.