Presents the H-principle, the Heisenberg modelling principle. General properties of the Heisenberg modelling procedure is developed. The theory is applied to principal component analysis and linear regression analysis. It is shown that the H-principle leads to PLS regression in case the task is linear regression analysis. The book contains different methods to find the dimensions of linear models, to carry out sensitivity analysis in latent structure models, variable selection methods and presentation of results from analysis.