Principal component analysis (PCA) is the most commonlyused chemometric technique. It is an unsupervised patternrecognition technique. PCA has found applications in chemistry,biology, medicine and economics. The present work attemptsto understand how PCA work and how can we interpretits results.
Partial least square (PLS) analysis is the most favourite tool in chemometrics to develop calibration models. PLS technique allows us to decipher even the complex systems by analysing all the variables instead of looking at them one at a time. PLS technique not only capture the maximum variation associated with predictor (i.e. spectra) and predicted (i.e. concentration) variables but also maximises the correlation between them. The present article describes the working scheme of the PLS algorithm. It also describes important technical details that need to be considered for developing a parsimonious and robust calibration model.
