Multivariate statistical analysis with experimental data

Abstract

Collected data from the sensors monitoring the environment in oil industry are various and raw, multivariate statistical analysis can turn these data into meaningful information. This paper would introduce some typical multivariate analysis methods, and investigate the data gathered in the Biota Guard exposed experiment by the means of some appropriate multivariate statistical analysis. Principal component analysis produces the principal components to represent the information of the multivariate in a reduced dimensional space; clustering analysis can group the observations of the multivariate into clusters in different ways; discriminant analysis can classifies new observations to existed clusters based on training data. These statistical analyses help us to understand the underlying information of the data from experiment and comparison of these analyses would distinguish the certain application of these methods in different situations and gives guidelines to further study.