ABSTRACT
Data analysis has long been recognized as an important source of competitive advantage in the biopharmaceutical industry. One can classify data analysis into three categories: overview of data, prediction, and classification. Multivariate data analysis (MVDA) tackles these three categories with a chemometric approach. The flexibility of multivariate methods has made them useful for the analysis and modeling of complicated and cumbersome data. These methods are increasingly used in a wide range of applications in biopharmaceutical development and manufacturing. These typically include process monitoring, early fault detection, quality control, and final product quality prediction (i.e., deriving relationships between process parameters and product quality attributes). Typical business benefits that can be achieved with
5.1 Introduction ....................................................................................................77 5.2 Multivariate Data Modeling Fundamentals .................................................... 78
5.2.1 Multivariate Modeling Basics ............................................................. 78 5.2.2 Training vs. Testing Datasets ..............................................................80 5.2.3 Data Verification and Preprocessing ..................................................80 5.2.4 Model Building ................................................................................... 81 5.2.5 Robustness Testing ............................................................................. 82 5.2.6 Maintaining Models ........................................................................... 82 5.2.7 Data Visualization and Multivariate Charts ....................................... 83 5.2.8 Documentation of Modeling ...............................................................84
