PLS Regression Calculator

Calculate a partial least squares (PLS) regression online, simply copy your data into the table above, select a metric dependent variable and one or more independent variables, and choose "Partial least squares" as the calculation method.

PLS Regression

Partial least squares (PLS) regression is a variant of linear regression that first compresses the independent variables into a small number of components and then predicts the dependent variable from these components. The components are chosen so that they capture as much of the covariance between the independent variables and the dependent variable as possible.

PLS regression is particularly useful when the independent variables are strongly correlated (multicollinearity) or when there are many predictors relative to the number of cases, situations in which ordinary least squares regression becomes unstable or cannot be computed at all.

Number of components

The key setting in a PLS regression is the number of components. The calculator suggests the number of components with the smallest prediction error, determined by 10-fold cross-validation, and you can adjust this value at any time to inspect simpler or more complex models.

Results

In addition to the model summary and the regression coefficients, the calculator reports the variance explained in X and Y per component, the variable importance in projection (VIP) scores, and the weights and loadings of each component. Predictors with a VIP score greater than 1 are usually considered important for the model.