Multivariate modelling of the pharmaceutical two-step process of wet granulation and tableting with multiblock partial least squares

Combining DI-ESI–MS and NMR datasets for metabolic profiling

“…More specifically, the CPCA-W algorithm (Westerhuis, Kourti, Macgregor 1998) was used to generate the MB-PCA model. MB-PLS with super-score deflation (Westerhuis, Coenegracht 1997) was used to generate the MB-PLS model. Both blocks were scaled to unit variance prior to modeling, and equal contribution of each block to the models (fairness) was ensured by further scaling each block by the square root of its variable count (Smilde, Westerhuis, de Jong 2003).…”

Multiblock redundancy analysis: interpretation tools and application in epidemiology

“…In other words, the focus in GCA-RT is not on recovering the total variance of the dataset to be explained as it is the case in mbRA. In addition, the first-order solution of GCA-RT is given by u (1) , the eigenvector of the matrix P Y P k P X k À Á , or alternatively by t (1) , the eigenvector of the matrix ( P k P X k P Y ), associated with the largest eigenvalue. Because GCA-RT involves the inversion of matrices (X 0 k X k ) and (Y 0 Y), it is sensitive to the multicollinearity in either X k or Y.…”

“…. The solution is given by v (1) , the eigenvector of (Y 0 XX 0 Y), or alternatively by w (1) the eigenvector of the matrix (X 0 YY 0 X), associated with the largest eigenvalue [2,18]. It follows that mbPLS is neither sensitive to multicollinearity in X k nor in Y, and leads then to a more stable solution.…”

“…The farms which have a risk profile for the disease are those for which the rates of seropositive sows and piglets are low (few antibody transmission from the sow to the piglets) and the rate of fattening pigs is high (high infection pressure). It turns out that the risk factors are associated with negative values on both global components t (1) and t (2) whereas protective factors are associated with positive values. In comparison with the block importance results, this graphical display also highlights that the explanatory variables from X 2 and X 4 have on the whole weak contributions to the first two global component inertia.…”

Multiblock modelling to assess the overall risk factors for a composite outcome

“…As the explanatory variables have been standardized, the total variance in each block is equal to the number of variables in this block. This motivates the block scaling in order to put the blocks on the same footing [30]. For this purpose, each of the ( K =4) explanatory block is accommodated with a scaling factor to set them at the same total variance 1/ K .…”