feature-set-comp

Compares various time-series feature sets on computational performance, within-set structure, and between-set relationships.

Repository Structure

The entire project folder structure and package loads are handled by the setup.R script. Please run this before any other scripts and each time you load up the project.

The main chunk of the repository is largely organised around the analysis/ folder which contains subfolders for each discrete topic of analysis:

• feature-calcs/ - Computation of time-series features
• comp-time/ - Benchmarking of feature set evaluation speed
• correlation/ - Pairwise feature-feature relationships across feature sets
• redundancy/ - Principal components analysis of within-set feature composition

The webscraping/ folder contains all the scripts necessary to automatically download and process the Empirical 1000 dataset used in this project.