pls

An algorithm worth exploring

pls is a Scala implementation of the partial least squares (PLS) algorithms described in Dayal and MacGregor’s Improved PLS Algorithms paper.

The project grew from an interest in regression methods and the relationships between predictor and response variables.

From a paper to an implementation

Implementing a statistical method makes its assumptions and representations concrete. The repository includes artificial datasets generated from known coefficients, along with examples of how those datasets were produced.

Those known relationships provide useful inputs for investigating the implementation’s behavior as the number of observations and predictors changes.

What this project explores

• Partial least squares algorithms.
• Statistical computing in Scala.
• Synthetic datasets with known underlying relationships.
• The practical work of turning an algorithm description into executable software.