A Tutorial to Developing Statistical Models for Predicting Disqualification Probability lected after the fitting of the model and N T denotes the number of observations in the test data set. THE DATA SETS The modeling approach developed is application independent, and it is demonstrated by using three data sets from different manufacturing industry applications. The case manufacturing processes are heat treatment of steel bars at Ovako Bar rolling mill, cheese production at Valio plant and strip rolling at Ruukki strip mill. All illustrations presented in this study are obtained from these case applications. The illustrations are calculated using test data sets which are not used in model fitting. In the first data set, the aim is to predict the disqualification probability related to mechanical properties of rolled steel bars after quenching and tempering. Modeling is done for two response vari- in three process stages in which different input variables are available so that the total number of models is six. In the third data set, the aim is to predict the disqualification probability related to the width of a hot rolled steel strip. Each strip is manufactured to a certain customer order and each customer has a requirement for the minimum width. Failing to fulfill the minimum width leads to scrapping or degradation of strips and is very expensive. The width is measured for each strip using a gauge that is located in the production line. Each mil- limeter of extra width means loss of material and costs for the rolling mill. The disqualifications are determined based on width excursion, which is a summary measure over the thousands of width measurements that are made from a single steel strip.