Based on the results of a comprehensive analysis made in the first part of the article, here we are handling the issue of developing prediction models on a large industrial dataset. The outcome of the first part is to ascertain, through a careful analysis, which of the process parameters will be selected as regressors and whom will be predictors. In this second part of the article they will be used in several models and the results will be assessed by their efficiency. Though formally distinct, these two parts should be read together, as they are mutually linked together. In practice, continuous exchange of information should be considered between these two parts.