experiment-design

Easily create and extend randomized experiment designs with the experiment-design library. Currently, it supports the creation of experiment designs using

• Random sampling

• Latin hypercube sampling

• Orthogonal sampling

Experiment variables can be discrete and continuous as well as correlated. experiment-design supports all these cases and aims to provide a space-filling designs with minimal correlation error.

Documentation Contents:

• Quickstart
  • Installation
  • Create and extend experiment designs
• Experiment Design via Orthogonal Sampling
  • Latin hypercube sampling
  • Orthogonal sampling
  • Why should you use orthogonal sampling?
  • Why choose experiment-design?
• Benchmarking some Latin hypercube sampling libraries
  • Results
  • Tested libraries
  • Methods and metrics
  • Conclusion
• API Reference
  • covariance_modification
  • optimize
  • orthogonal_sampling
  • random_sampling
  • scorers
  • variable
• Citing
• Contributing