/progml

Working through Programming Machine Learning

Primary LanguageRacket

Working through "Programming Machine Learning"

Some in Racket, some in Python.

Notes on performance

Untyped linear regression (With rendering disabled)

  • direct runs ~6.3s
  • after raco make ~5.8s

Typed linear regression (With rendering disabled)

  • direct runs ~3.5s
  • after raco make ~1.4s
  • removing the plot import - ~550ms. which 3.6x slower than Python using numpy. Not bad.