Skip to content
forked from yangkky/gpmodel

Functions and classes for doing Gaussian process models of proteins

Notifications You must be signed in to change notification settings

DNA2RNA/gpmodel

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GPModel

Gaussian process regression and classification with NumPy and SciPy.

Getting started

There are currently three families of models: classification, regression, and regression with lasso for feature selection. There are also many kernels already implemented, including squared exponential, Matern, linear, and polynomial.

To build a model, first choose a kernel. For example,

ke = gpkernel.PolynomialKernel(3)

instantiates a cubic kernel.

Next, instantiate the model with this kernel. In this case, we'll use a regression model:

mo = gpmodel.GPRegressor(ke)

We can fit the model by passing it training data as NumPy arrays. In the Gaussian process context, fitting the model means choosing kernel hyperparameters (and the noise hyperparameter for regression models) that maximizes the log marginal likelihood.

_ = mo.fit(X, y)

We can also use the model to make predictions:

means, cov = mo.predict(X_test)

This returns the full predictive distribution as a vector of means and the covariance matrix.

Further Reading

This package implements algorithms from chapters 2, 3, and 5 of Rasmussen and William's Gaussian Processes for Machine Learning.

About

Functions and classes for doing Gaussian process models of proteins

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%