PLS Partial Least Squares Projections to Latent Structures

PLS Partial Least Squares Projections to Latent Structures 

The PLS method was developed by Wold.[7] It has now been applied to problem solving in so many different areas of science and technology that it is not possible to give even a brief comment to them all. We shall see that PLS is a very useful tool, also in organic synthesis. 

PLS is a modelling and computational method, by which quantitative relations can be established between blocks of variables, e.g. a block of descriptor data for a series of reaction systems (X block) and a block of response data measured on these systems (Y block). By the quantitative relation between the blocks, it is possible to enter data for a new system to the X block and make predictions of the expected responses. For example, if a reaction has been run in a series of solvents, we can use a PLS model to relate the properties of the solvents to the observed optimum conditions in these solvents. By subsequently entering the property descriptors of a new solvent to the PLS model it is possible to predict the optimum conditions of the reaction in the new solvent. For this to be efficient, it is necessary that the solvents used to determine the PLS model have a sufficient spread in their properties. To ensure this, a design in the principal properties is most useful. 

The model can also, conversely, be used to rind which conditions X give a certain desired Y-profile. 

One great advantage of the method is that the results can be evaluated graphical by different plots. In most cases, visual interpretations of the plot are sufficient to obtain a good understanding of different relations between the variables. 

---

E Johansson and M Cocchi 1993. PLS - Partial Least-squares Projections to Latent Structures. In binyi H (Editor) 3D QSAR in Drug Design. Leiden, ESCOM, pp. 523-550. 

PLS Partial least squares projection to latent structures... 

Wold, S., Johansson, E., Cocchi, M., PLS - Partial least-squares projections to latent structures, in 3D QSAR in Drug Design. Kubinyi, H. (ed.). ESCOM, Leiden, 1993, pp. 523-... 

Wold S, Johansson E, Cocchi M. PLS-partial least squares projections to latent structures. In Kubinyi H, editor. 3D-QSAR in drug design Theory, methods and applications. Leiden ESCOM, 1993. p. 523-50. 

MLR - multiple linear regression, PLS - partial least squares projection to latent structures, NN - neural network. 

S. Wold, E. Johansson, and M. Cocchi, in 3D Q5AR in Drug Design. Theory Methods and Applications, H. Kubinyi, Ed., ESCOM, Leiden, 1993, pp. 523-550. PLS—Partial Least-Squares Projections to Latent Structures. 

---