The French-style of data analysis developed by Jean-Paul Benzécri, data analysis à la Française was based on the philosophical assumption which acknowledges the development of models to fit the data, instead of the rejection of hypotheses based on the lack of fit. Visual presentations make it easier to intuitively understand the data structure and “to stabilize and memorize” the forms of the relationships and further interpretation. The different scientific communities do not agree with the same definitions for the heterogeneity grouping/discrimination taxonomies. The assumption of the latent data structure is homogenous is often unrealistic. On the measurement model level, sequential segmentation strategies, usually fail to identify groups of units with distinctive inner path model estimates. The solution proposed by researchers is based on other segmentation approaches; studies show that these techniques continue to suffer from deficiencies related for example with the types of heterogeneity covered, shapes of clusters or distributional assumptions. The proposed method, KN-PLS is appropriate for complex type of shapes of data creating well defined latent classes.