explained variance by each perdictor variable in plsregress

10 次查看（过去 30 天）

Simon Lind 2018-10-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/421551-explained-variance-by-each-perdictor-variable-in-plsregress

评论： Simon Lind 2021-8-10

I'm using plsregress to generate a model to reproduce my variable as a function of a set of 10 predictors. I need to know the explained variance by each predictor (to select the most important ones). PCTVAR gives some information but it doesn't say which one explains the highest proportion of the variance.

If X is my matrix of predictors, and I change the order of the variables: X = [var1 var2 var3 ... varn] to X = [var3 var2 var1 ... varn], the results in PCTVAR are exactly the same.

thank you

2 个评论
显示无隐藏无

Jokin Ezenarro 2021-8-7

I think that is not calculated in a PLS model, as the predictors are inside the LVs. You could use the regression vector shown in the BETA matrix, the higher the coefficient, the bigger the importance if the predictors are in the same scale.

Simon Lind 2021-8-10