Proc PLS - interpretation of results

[Olanike]

Hello, I am running a Proc PLS to find the correlation between my dependent and independent variables, e.g Yield vs MinT MaxT. I am confused as to which  of the tables tells me the percentage of my variables (min T and Max T combined and separately) that explained yield.  My model was fine and below are some of my results.

Please how do I get the percentage of yield explained by a combination of  MinT and MaxT

The PLS Procedure
SC=DRY
Percent Variation Accounted for by Partial Least Squares Factors
Number of Extracted Factors	Model Effects	Dependent Variables
Current	Total	Current	Total
1	40.0228	40.0228	4.749	4.749
2	59.9772	100	0.0097	4.7587

SC=DRY
Model Effect Loadings
Number of Extracted Factors	MinT	MaxT
1	0.583192	-0.81234
2	0.778905	0.627142
Model Effect Weights
Number of Extracted Factors	MinT	MaxT	Inner Regression Coefficients
1	0.628099	-0.78009	0.243576
2	0.778905	0.627142	0.008972
Dependent Variable Weights
Number of Extracted Factors	Wheat_Y
1	1
2	1

[PaigeMiller · Accepted solution]

PLS creates dimensions (sometimes called "factors" or "latent factors") which are used to predict Y. So, dimension 1 (which consists of a weighted combination of both minT and maxT) explains 4.749 percent of the Y variability, and dimension 2 (which consists of a weighted combination of both minT and maxT) explains another 0.0097% of the variability of Y. So I think the answer to your exact question "Please how do I get the percentage of yield explained by a combination of  MinT and MaxT" is given by this number, but you have to decide if you want to use one dimension, or two dimensions. 

You can obtain information that a certain percent of the variability of minT is used in dimension 1, and additional percent of variability of minT is used in dimension 2 (same is possible for maxT). You would add the VARSS option to the PROC PLS statement.

https://documentation.sas.com/?cdcId=pgmmvacdc&cdcVersion=9.4&docsetId=statug&docsetTarget=statug_pl...

[Olanike]

Thank you for the response.