pyRSM is an object oriented framework to create and use response surface approximations or metamodels. Such approximations are useful in different fields and are particularly useful in design and/or optimization were they can be used in liue of expensive function analyzes.
Different design of experiments (DOE) are included in the framework to sample a given design space
Halton sequences
Hammersley sequences
Latin Square sampling
Monte Carlo sampling
Random sampling
Taguchi sampling
DOEs sampling on example 2D design space¶
Different state of art response surface methodologies (RSM) are currently available in the framework.
Radial Basis Functions (RBF)
Multi Adaptive Regression Splines (MARS)
Least-Squares Support Vector Machine (LSSVM)
Gaussian Process Regression (GPR or Kriging)
Example RBF response prediction of 1D function¶
Example LSSVM response prediction of 1D function with noise¶