I know the answer to this question, so maybe this should instead be posted in the "wish list," though maybe there is a work-around: I have complex datasets in which I have many Y's (analytes) and many X's (environmental data). Pretty much the only JMP Pro platform capable of making multi-Y/multi-X models is partial least squares and generalized regression (I think). After sitting through a really excellent webinar by @kemal_oflus on neural networks (which I use for modeling single Y responses with many X's), I am wondering if there is any way to tap into the power of JMP's neural network platform to model multivariate data. I am talking about things like assemblages of organisms or genomic datasets for which you want to know the most influential environmental drivers of variation. I guess Gen-reg and PLS are fine for handling these sorts of datasets in all honesty, but just out of curiosity could a neural network model be used to yield a multi-Y prediction? For instance, maybe you want to know the environmental conditions that are most suitable for two species of coral. I guess you'd simply build a model for each coral species and compare them, though this could get cumbersome with ecosystems with many different inhabitants!
Anderson B. Mayfield
