pyttb.gcp.handles

Implementation of the different function and gradient handles for GCP OPT.

class pyttb.gcp.handles.Objectives(value)

Bases: Enum

Valid objective functions for GCP.

GAUSSIAN = 0

BERNOULLI_ODDS = 1

BERNOULLI_LOGIT = 2

POISSON = 3

POISSON_LOG = 4

RAYLEIGH = 5

GAMMA = 6

HUBER = 7

NEGATIVE_BINOMIAL = 8

BETA = 9

pyttb.gcp.handles.gaussian(data: ndarray, model: ndarray) → ndarray

Return objective function for gaussian distributions.

pyttb.gcp.handles.gaussian_grad(data: ndarray, model: ndarray) → ndarray

Return gradient function for gaussian distributions.

pyttb.gcp.handles.bernoulli_odds(data: ndarray, model: ndarray) → ndarray

Return objective function for bernoulli distributions.

pyttb.gcp.handles.bernoulli_odds_grad(data: ndarray, model: ndarray) → ndarray

Return gradient function for bernoulli distributions.

pyttb.gcp.handles.bernoulli_logit(data: ndarray, model: ndarray) → ndarray

Return objective function for bernoulli logit distributions.

pyttb.gcp.handles.bernoulli_logit_grad(data: ndarray, model: ndarray) → ndarray

Return gradient function for bernoulli logit distributions.

pyttb.gcp.handles.poisson(data: ndarray, model: ndarray) → ndarray

Return objective function for poisson distributions.

pyttb.gcp.handles.poisson_grad(data: ndarray, model: ndarray) → ndarray

Return gradient function for poisson distributions.

pyttb.gcp.handles.poisson_log(data: ndarray, model: ndarray) → ndarray

Return objective function for log poisson distributions.

pyttb.gcp.handles.poisson_log_grad(data: ndarray, model: ndarray) → ndarray

Return gradient function for log poisson distributions.

pyttb.gcp.handles.rayleigh(data: ndarray, model: ndarray) → ndarray

Return objective function for rayleigh distributions.

pyttb.gcp.handles.rayleigh_grad(data: ndarray, model: ndarray) → ndarray

Return gradient function for rayleigh distributions.

pyttb.gcp.handles.gamma(data: ndarray, model: ndarray) → ndarray

Return objective function for gamma distributions.

pyttb.gcp.handles.gamma_grad(data: ndarray, model: ndarray) → ndarray

Return gradient function for gamma distributions.

pyttb.gcp.handles.huber(data: tensor, model: tensor, threshold: float) → ndarray

Return objective function for huber loss.

pyttb.gcp.handles.huber_grad(data: tensor, model: tensor, threshold: float) → ndarray

Return gradient function for huber loss.

pyttb.gcp.handles.negative_binomial(data: ndarray, model: ndarray, num_trials: float) → ndarray

Return objective function for negative binomial distributions.

pyttb.gcp.handles.negative_binomial_grad(data: ndarray, model: ndarray, num_trials: float) → ndarray

Return gradient function for negative binomial distributions.

pyttb.gcp.handles.beta(data: ndarray, model: ndarray, b: float) → ndarray

Return objective function for beta distributions.

pyttb.gcp.handles.beta_grad(data: ndarray, model: ndarray, b: float) → ndarray

Return gradient function for beta distributions.