Heirarchical Partioning of Variance

I'm trying to develop a multilple linear regression model to predict phytoplankton biomass in lake based on a set of envrionemntal variables (temperature, wind, oxygen, nutrients, etc). There's a lot of multi-collinearities between predictors making it difficult to find the "best" multiple linear regression model. I would like to use heirarchical variance paritioning to identify the amount of variance explained by each predictor indepedent of all others to find the best combination of predictors. In R-suite there is a "Varpart" function to do this. Is there any similiar package in Matlab that i could use?

[Ive J]

no AFAIK. There is no such implementation of redundancy analysis ordination MATLAB. I guess one could develop something based on manova1 though. R is best you got for this purpose.

[Micah Swann]

thank you for the suggestion

Answer 1

[Sarthak]

Hi Micah,

As per my understanding, to perform multiple linear regression you could use the Regression Learner App or the ‘regress’ function.

For the analysis of variance for linear regression models, you can use the ‘anova’ function.

You can also have a look at the Fathom Toolbox for MATLAB in File Exchange with some related function implementations.

Attaching all documentation and function definition links for your reference.

• https://www.mathworks.com/help/stats/multiple-linear-regression-1.html
• https://www.mathworks.com/help/stats/regress.html
• https://www.mathworks.com/help/stats/linearmodel.anova.html
• https://www.mathworks.com/matlabcentral/fileexchange/68518-fathom-toolbox-for-matlab
• https://www.mathworks.com/help/stats/hierarchical-clustering.html

[Ive J]

OP is asking about RDA and variance partitioning and not multiple linear regression or anova (overall variance). MATLAB does not have any built-in functions for RDA. This response is misleading.