Creating Predictive Models with the Results of Multivariate Analyses

So far, we've discussed the results of multivariate analysis in terms of correlations and associations, where the goal is to understand and explain the relationships. The next step is to use these results to create models that can predict the outcome variable given a new set of predictor measurements.

It may sound difficult, but it's really not. If you've run your multivariate analysis and got a good, sound result, then you have all you need to be able to make a good, sound prediction.

So far, so good, but notice that I said the result of finding the significant predictor variables in a multivariate analysis will be a good, sound model. I didn't say it would be the best or optimum model. Because the requirements of the multivariate technique are different when finding relationships to making predictions using them, your modeling method needs to be adjusted to account for that.

When seeking relationships...