Base Model Class#

class regmmd.models.base_model.EstimationModel[source]#

Bases: ABC

abstractmethod log_prob(x)[source]#

Evaluates to the log likelihood at the values x.

Parameters:

x (ndarray of shape (n_samples, n_features), the points to be evaluated)

Return type:

ndarray[tuple[Any, …], dtype[_ScalarT]]

abstractmethod sample_n(n)[source]#

Generate n samples of the distribution with the initialized parameters of the distribution.

Parameters:

n (int, How many samples to generate)

Return type:

ndarray[tuple[Any, …], dtype[_ScalarT]]

abstractmethod score(x)[source]#

Evaluates to the gradient of the log likelihood with respect to the parameters at the values x.

Parameters:

x (ndarray of shape (n_samples, n_features), the points to be evaluated)

Return type:

ndarray[tuple[Any, …], dtype[_ScalarT]]

abstractmethod update(par_v)[source]#

Update the model with new parameters

Parameters:

par_v (float, variable parameters)

Return type:

None

class regmmd.models.base_model.RegressionModel[source]#

Bases: EstimationModel

abstractmethod predict(X)[source]#

Computes the mean of Y given X and the current parameters of the model

Parameters:

X (np.array)

Return type:

ndarray[tuple[Any, …], dtype[_ScalarT]]

abstractmethod sample_n(n, mu_given_x)[source]#

Generate n samples of the distribution with the initialized parameters of the distribution and the conditional mean of the covariates

Parameters:

  • n (int, How many samples to generate)

  • mu_given_x (np.array, the covariates)

Return type:

ndarray[tuple[Any, …], dtype[_ScalarT]]