Comparing the two methods proposed (MPRM) and (MPLSKURSD) with the modalities of partial least squares (PLS) fortified


The emergence of multi-linear full problem in the explanatory variables to model multi Alanhaddaralkhti variables makes it difficult to apply the classical methods such as the method (ols) because they give inaccurate results and to address such a problem using other methods such as least-squares District (pls) .ala that this method be sensitive towards anomalous values, if any, in the data set, so it is advisable to resort to methods such as fortified (PRM) and (PLSKURSD). In this research will be to use the two methods above in addition to (pls) will be compared with the methods proposed (MPRM) and (MPLSKURSD) by simulation through two experiments first experiment relied on several types of anomalous values of data and different rates of pollution and volumes of samples and the dimensions of different variables and adopted a second on a comparison between the methods when the error is distributed naturally distributed in addition to other distributions. Terminology: partial least squares, gay, simpls algorithm, broker multivariate, covariance matrix fortified, kurtosis, projections, segmentation singular value