A methodology to detect outliers/inliers in prediction with PLS

A study of the homogeneity of the data should be performed in order to guarantee the detection of outliers and inliers in prediction with a PLS model. For this reason, we decided to develop an automatic methodology, with a possibility for visual checking, to detect these objects. This methodology consists of three steps. First, the objects are mapped from an n-dimensional space to a 2-dimensional space using Sammonʹs mapping. Then, clusters in the calibration space are detected using a density-based method, and finally, the convex hull method is applied to each cluster in order to detect outliers/inliers in new samples. Several case studies were carried out with this methodology. The results obtained show that the combination of these three different techniques makes the detection of outliers and inliers for prediction easier and more accurate than classical methods.