pls_regression.sas

%let path=your/path/here; 
libname multivariate "&path"; 
ods graphics on;

data concrete; 
    set multivariate.concrete; 
run;

/* 

    Partial Least Squares Regression

    PLS has been implemented using the NIPALS algorithm. 
    As the number of observations is small (n=103), we can use
    leave-one-out cross-validation.
    
    Water increases workability at the expense of compressive strength.
    Fly ash and concrete both increase compressive strength.
    Fine and coarse aggregates have an inverse relationship with one
    another. Plasticizer has minimal effects on the Y matrix.
*/ 

proc pls data = concrete method = pls(algorithm=nipals) 
    cv=one plot=(vip xyscores xscores parmprofiles dmod); 
    model slump--compressive_strength = cement--fine_aggregate; 
run;