Examples

For a thorough tutorial of the methods introduced in scikit-matter, we suggest you check out the pedagogic notebooks in our companion project kernel-tutorials.

PCovR and KernelPCovR

• Construct a PCovR Map
• Construct a Kernel PCovR Map
  • Note on KernelPCovR for Atoms, Molecules, and Structures
• The Benefits of Kernel PCovR for the WHO Dataset
  • Load the Dataset
    • Scale and Center the Features and Targets
  • Train the Different Linear DR Techniques
    • PCovR
    • PCA
  • Train the Different Kernel DR Techniques
    • Select Kernel Hyperparameters
    • KPCovR
    • KPCA
    • Correlation of the different variables with the KPCovR axes
  • Plot Our Results
• Choosing Different Regressors for PCovR
  • Use the default regressor in PCovR
  • Use a fitted regressor
  • Use a pre-predicted y
  • Comparing Results
• The Importance of Data Scaling in PCovR / KernelPCovR

Feature and Sample Selection

• PCovR-Inspired Feature Selection
  • Feature Selection with CUR + PCovR
    • Non-iterative feature selection with CUR + PCovR
  • Feature Selection with FPS + PCovR
• Feature Selection on the WHO Dataset
  • Load the Dataset
  • Scale and Center the Features and Targets
  • Provide an estimated target for the feature selector
  • Compute the Selections for Each Selector Type
    • PCov-CUR
    • PCov-FPS
    • CUR
    • FPS
    • (For Comparison) Recurisive Feature Addition
  • Plot our Results
• Using scikit-matter selectors with scikit-learn pipelines
  • Simple integration of scikit-matter selectors
  • Stacking selectors one after another

Orthogonal Regression

• Regression with orthogonal projector/matrices

Feature Reconstruction Measures

• Global Feature Reconstruction Error (GFRE) and Distortion (GFRD)
• Pointwise GFRE applied on RKHS features
• Pointwise Local Reconstruction Error