*Benchmarking best practice*
Ruthlessly optimize your method on one dataset, or maybe three.
As for other people's methods, read the paper title and glance at the figures to know what they're doing and then quickly implement it yourself. Set parameters to some unreasonable default and off you go!

space without doing real work for it. Written by first-time coders, no unit or integration tests, complimentary mismatches between code and documentation. All methods work on our example data. If you can install them (good luck). We like to reinvent. Everything before Python was useless anyway.(2/2)

Announcing the ReInvPy project!

Basic statistics and maths concepts but written in Python and renamed so you can impress your PI. Lots of user-defined tuning parameters so that all methods forever remain moving targets in method comparisons and benchmarks. Occupies places in idea space... (1/2)