why is the data different after denormalization

3 次查看（过去 30 天）

Kia 2024-9-29

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2156200-why-is-the-data-different-after-denormalization

编辑： John D'Errico 2024-9-29

I used ARESLAB on matlab and obtained an R2 for train and test greater than 99.0% with a MAE and MSE less than 10-3. Initially, I had normalized the database with Zcore, so after getting these results, I wanted to plot them so I wanted to denormalize the data. But the results obtained sometimes gave predictive variables different from those measured, even though the R2 was good and the errors were relatively good.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

John D'Errico 2024-9-29

编辑：John D'Errico 2024-9-29

Without the data, and seeing EXACTLY what you did, this is difficult to know. Did you just make a mistake? Gosh, that never happens. ;-) I mean, it never happens that someone mistakenly uses the wrong information in the wrong place, and then claims something strange happened on Answers. Can we possibly know that is not what you did? And even at that, R^2 is just a number, one that is easily fooled. Placing too much reliance on that one number often causes problems.

So if you want help, then show what you did. EXACTLY. Provide your data. Then someone can show what happened. Anything else is just will just lead to wild guesses from anyone who tries to answer.

请先登录，再进行评论。

请先登录，再回答此问题。