RoadMap

Description of planned features and schedule

Introduction

This page lays out the new features I plan on including in the toolkit, and their priority (may not necessarily be up to date -- also see Issues page for marked enhancements).

Target Release 2.0:

Highest priority (required for 2.0):

• Align interfaces for estimators for discrete and continuous-valued data - see Issue 66
• Add a new console panel to the AutoAnalyser for longer output for more complex calculations - see Issue 61
• Add support for multivariate calculations to the AutoAnalyser - see Issue 78

Medium priority:

• Implement dynamic/serial correlation correction for Gaussian estimators (at minimum by sub-sampling, more investigation required for more advanced methods) and discrete estimators
• Implement analytic forms for null surrogate distributions for KSG estimators in statistical significance calculations (requires full investigation here)

Expansions of functionality (lower priority)

• Implement Partial Information Decomposition estimators, pulling in our previous framework from JPID and adding implementation of Conor's Pointwise PID measures
• KSG estimators for periodic variables, e.g. for information measures on phase values, first merging in the PR from Oscar for MI, then engineering this into conditional MI as well.
• Phase transfer entropy - add a wrapper class that determines the phases (via a Hilbert transform, as done in Fraschini's code) then passes into either a discretised or KSG estimator (turning on periodic boundary conditions)
• NSB estimator for entropies of discrete data