Understanding the Impact of the Properties of Raw Materials Onto Product Quality Using JR-PLS

Multiple proposals are found in literature to analyze the impact of raw materials onto product using data analysis. From the methods proposed, the L-PLS is the only method that not only considers the ratio in which ingredients are mixed, but also the physical properties of the lots of raw material. This method however is limited to analyze data from materials of the same nature such that all the physical properties apply to all materials. This is not the case when a product includes materials that need to undergo different physcial/chemical tests due to theiir nature. Modeing this scenario using the L-PLS would include blocks of missing data with non-overlapping rows or columns.

A new method is presented to handle this situations, the Joint-R PLS (JRPLS) combines the JYPLS and the LPLS methods providing a plattform to model the impact of the physical/chemical properties of multiple ingredients of different natures onto the final quality attributes of the product. The use and advantage of the model is shown in the troubleshooting of multiple scenarios where a complex relation of physical/chemical properties across multiple ingredients was identified as the root-cause of a situation being investigated.