Hi, everyone,

We are developing a pedotransfer function (PTF) to estimate lime requirement (LR) of acid soils. The data are intended to be submitted to Stepwise Regression Analysis. We now have a number of soil properties closely correlated to soil LR. Has anyone known any proper method to define groups of predictors for PTFs ? Can we proceed with all of the predictors we have, or just work on some combinations of predictors ? Thanks a lot.