easypheno.model._bayesian_linreg

Module Contents

Classes

Bayes

Implementation of a class for Bayesian linear regression.
class easypheno.model._bayesian_linreg.Bayes(task, encoding=None, iterations=100, warmup=10)

Bases: easypheno.model._param_free_base_model.ParamFreeBaseModel

Implementation of a class for Bayesian linear regression.

Attributes

Inherited attributes

See ParamFreeBaseModel for more information on the attributes.

Additional attributes

  • mu (np.array): intercept

  • beta (np.array): effect size

  • iterations (int): MCMC sampler iterations

  • warmup (int): number of discarded MCMC warmup iterations

Parameters

  • task (str) –

  • encoding (str) –

  • iterations (int) –

  • warmup (int) –

standard_encoding = 012
possible_encodings = ['101']
abstract probability_model(self, X, y)

Probability model that needs to be implemented by each child model.

Parameters

  • X (torch.Tensor) – feature matrix

  • y (torch.Tensor) – target vector

fit(self, X, y)

Implementation of fit function for Bayesian linear regression.

See ParamFreeBaseModel for more information.

Parameters

  • X (numpy.array) –

  • y (numpy.array) –

Return type

numpy.array

predict(self, X_in)

Implementation of predict function for Bayesian linear regression.

See ParamFreeBaseModel for more information.

Parameters

X_in (numpy.array) –

Return type

numpy.array