PLS Model Building

PLS analysis is carried out on input data to build a multiple linear regression model that can then be applied to other data sets. Both training set and testing set are required as input, these can be created using the Partitioning or Row Splitter KNIME nodes. Input data must contain 2 or more columns of numerical data, for the independent variables (X) and the dependent variable (Y).

Backend implementation

utilities/canvasPLS
canvasPLS is used to implement this node.

Input ports

1. Type: Table
   Training set variables
   Numerical data of the training set variables
2. Type: Table
   Test set variables
   Numerical data of the test set variables (identical Variable columns as defined for the Training set)

Output ports

1. Type: Table
   PLS Model and statistics for each factor
   PLS Model and statics for each factor
2. Type: Table
   Factor loadings
   Factor loadings
3. Type: Table
   Plot data
   Plot data showing observed and predicted values of Y for both the training and test sets.