: Development of graphical interface based on exploratory analysis for multivariate chemical data
Chemometrics. Classification. Multivariate analysis. GUI.
This dissertation aims to demonstrate application of supervised and exploratory classification algorithms in chemical data, showing the potential of this combination. For this, three research articles are available. Initially, two applications based on infrared spectroscopy and supervised classification are addressed. Finally, a computational tool based on exploratory analysis is proposed as a technological innovation for data analysis. The first article consists of the application of chemometric tools as na alternative method of screening for breast cancer, aiming at rapid, low-cost, minimally invasive and statistically reliable detection when compared to the mammography method, which is now established as the standard method for screening and detection. The second article demonstrates the application of spectroscopic and supervised classification methods in HIV-infected patients, aiming at their rapid detection through blood plasma at a low operating cost. The third article presents a graphical user interface (GUI) built as a user-friendly, iterative, educational and scientific tool with a strong visual appeal. This interface uses several clustering methods, based on hierarchical group analysis (HCA), as well as principal component analysis (PCA) for exploratory data study, focusing on first-order spectroscopic data. Furthermore, the interface can be used for data processing and cluster estimation, presenting itself as a potential tool for multivariate chemical data analysis. The results presented here are useful and innovative, serving as theoretical and practical support for future chemometric applications.