ROADMAP

[duodenum96]

The current roadmap for IntrinsicTimescales.jl:

• Add variable exponent option and option for constraints for FOOOF style fitting
• Make Turing.jl an extension package. This should further reduce precompilation time.
• Better diagnostic plots for fitting based INT estimations (:tau and :knee)
• Reduce TTFX problem with PrecompileTools.jl
• Finish the python frontend (taking inspiration from https://github.com/SciML/diffeqpy)
• Complete the theory section in the documentation
• I think Bayesian methods deserve better visualization utilities
• Batch processing option if the user wants to process multiple subjects/channels without a for-loop. Possible parallelism here.
• There are a number of possible performance improvements in the repository (such as using FFT plans, static arrays in certain places). Handle them.
• As a more long-term goal: extending approximate Bayesian computation to hierarchical approximate Bayesian computation for multiple subjects