Multivariate Data Analysis

Different chapters within this book and the literature prove the ToF-SIMS as a powerful surface analysis tool. However, one large drawback is that data analysis can be challenging and is very time-consuming compared to the measurement time and to other analysis techniques. Often only a small subset of data is analysed, e.g. a peak list of a few selected signals. Nowadays, data sets often get more complex, as the instruments are highly automated and even three-dimensional (3D) data sets can be easily acquired. In addition, the combination of different analysis techniques within the instrument or different surface analysis techniques leads to more complex data sets with high amounts of hidden meaningful information. To aid the analysis of these large data sets, multivariate data analysis (MVA) has for many years proven to be a powerful tool. For the successful application of MVA methods, the right data pretreatment and understanding of the method are crucial. The basics of the usage of MVA leading to meaningful results are summarized in this chapter. Besides, particular emphasis will be on developments within the last decade: the identification of small differences in spectra (e.g. for forensic analysis), concepts to simplify data interpretation of MVA results and applications in machine learning.

Перевод пока недоступен