PLSRegressor.jl

A Partial Least Squares Regressor package. Contains PLS1, PLS2 and Kernel PLS2 NIPALS algorithms. Can be used mainly for regression. However, for classification task, binarizing targets and then obtaining multiple targets, you can apply KPLS.

PackageEvaluatorBuild Status

Install

Pkg.add("PLSRegressor")

Using

using PLSRegressor

Examples

using PLSRegressor

# learning a single target
X_train        = [1 2; 2 4; 4 6.0]
Y_train        = [4; 6; 8.0]
X_test         = [6 8; 8 10; 10 12.0]
Y_test         = [10; 12; 14.0]

model          = PLSRegressor.fit(X_train,Y_train,nfactors=2)
Y_pred         = PLSRegressor.predict(model,X_test)

print("[PLS1] mae error : $(mean(abs.(Y_test .- Y_pred)))")


# learning multiple targets
X_train        = [1 2; 2 4; 4 6.0]
Y_train        = [2 4;4 6;6 8.0]
X_test         = [6 8; 8 10; 10 12.0]
Y_test         = [8 10; 10 12; 12 14.0]

model          = PLSRegressor.fit(X_train,Y_train,nfactors=2)
Y_pred         = PLSRegressor.predict(model,X_test)

print("[PLS2] mae error : $(mean(abs.(Y_test .- Y_pred)))")

# nonlinear learning with multiple targets
model          = PLSRegressor.fit(X_train,Y_train,nfactors=2,kernel="rbf",width=0.1)
Y_pred         = PLSRegressor.predict(model,X_test)

print("[KPLS] mae error : $(mean(abs.(Y_test .- Y_pred)))")


# if you want to save your model
PLSRegressor.save(model,filename=joinpath(homedir(),"pls_model.jld"))

# if you want to load back your model
model = PLSRegressor.load(filename=joinpath(homedir(),"pls_model.jld"))

What is Implemented

  • A fast linear algorithm for single targets (PLS1 - NIPALS)
  • A linear algorithm for multiple targets (PLS2 - NIPALS)
  • A non linear algorithm for multiple targets (Kernel PLS2 - NIPALS)

What is Upcoming

  • Bagging for Kernel PLS
  • An automatic validation inside fit function

Method Description

  • PLSRegressor.fit - learns from input data and its related single target

    • X::Matrix{:<AbstractFloat} - A matrix that columns are the features and rows are the samples
    • Y::Vector{:<AbstractFloat} - A vector with float values.
    • nfactors::Int = 10 - The number of latent variables to explain the data.
    • copydata::Bool = true - If you want to use the same input matrix or a copy.
    • centralize::Bool = true - If you want to z-score columns. Recommended if not z-scored yet.
    • kernel::AbstractString = "rbf" - use a non linear kernel.
    • width::AbstractFloat = 1.0 - If you want to z-score columns. Recommended if not z-scored yet.

  • PLSRegressor.transform - predicts using the learnt model extracted from fit.

    • model::PLSRegressor.Model - A PLS model learnt from fit.
    • X::Matrix{:<AbstractFloat} - A matrix that columns are the features and rows are the samples.
    • copydata::Bool = true - If you want to use the same input matrix or a copy.

References

  • PLS1 and PLS2 based on

  • A Kernel PLS2 based on

  • NIPALS: Nonlinear Iterative Partial Least Squares

    • Wold, H. (1966). Estimation of principal components and related models by iterative least squares. In P.R. Krishnaiaah (Ed.). Multivariate Analysis. (pp.391-420) New York: Academic Press.

  • SIMPLS: more efficient, optimal result

    • Supports multivariate Y
    • De Jong, S., 1993. SIMPLS: an alternative approach to partial least squares regression. Chemometrics and Intelligent Laboratory Systems, 18: 251– 263

License

The PLSRegressor.jl is free software: you can redistribute it and/or modify it under the terms of the MIT "Expat" License. A copy of this license is provided in LICENSE.md