def run(self, data, upc_sequence, resources):
"""
'data' is a 4D array (nobservations, nchoices, nvariables, number of nests).
'resources' can contain an entry 'starting_values' which is a dictionary where keys are
coefficient names and values are either starting values, or tuples with first element being
the starting value and second element being boolean determining if the coefficient
should be estimated (True) or not (False). Scaling parameters in the 'starting_values'
dictionary should start with '__logsum_' followed by the nest number.
'resources' can contain an entry 'bfgs_approximate_second_derivative' (logical) which determines if
the inverse Hessian is computed by a numerical approximation of the second derivative
(using finite difference method), instead of taking it out of the scipy BFGS function.
'resources' can contain an entry 'bfgs_epsilon' which is the step size for approximation
of the first derivative.
"""
return EstimationProcedure.run_dcm(self, data, upc_sequence, resources)
def run(self, data, upc_sequence, resources):
"""
'data' is a 4D array (nobservations, nchoices, nvariables, number of nests).
'resources' can contain an entry 'starting_values' which is a dictionary where keys are
coefficient names and values are either starting values, or tuples with first element being
the starting value and second element being boolean determining if the coefficient
should be estimated (True) or not (False). Scaling parameters in the 'starting_values'
dictionary should start with '__logsum_' followed by the nest number.
'resources' can contain an entry 'bfgs_approximate_second_derivative' (logical) which determines if
the inverse Hessian is computed by a numerical approximation of the second derivative
(using finite difference method), instead of taking it out of the scipy BFGS function.
'resources' can contain an entry 'bfgs_epsilon' which is the step size for approximation
of the first derivative.
"""
return EstimationProcedure.run_dcm(self, data, upc_sequence, resources)
def run(self, data, upc_sequence, resources):
"""
'data' is of shape (nobservations, nchoices, nvariables).
"""
return EstimationProcedure.run_dcm(self, data, upc_sequence, resources)
def mnl_loglikelihood(self, data, b, depm):
return EstimationProcedure.dcm_loglikelihood(self, data, b, depm)
def loglikelihood(self, beta, data, depm):
return EstimationProcedure.dcm_loglikelihood(self, data, beta, depm)
def run(self, data, upc_sequence, resources):
"""
'data' is of shape (nobservations, nchoices, nvariables).
"""
return EstimationProcedure.run_dcm(self, data, upc_sequence, resources)
