Fault Detection and Classification (FDC) is one of the major application areas for multivariate data analysis. Umetrics brings a unique and robust method of addressing the FDC problem. Using batch analysis techniques, Umetrics is able to not only detect faults in the step by step evolution of the recipe but also monitor wafer to wafer variation. This double-layer analysis gives excellent insight into the variation occurring in a process step as well as the entire recipe.