def calc_errorcomponent(self, variance_norm, variance_halfnorm,
vertex, size, seed):
"""
The method returns the contribution of the error component in the
calculation of the predicted value for the different choices.
Inputs:
variance_norm - numeric value (variance of the normal portion of error)
variance_halfnorm - numeric value (variance of the half normal portion of
error)
vertex - string (the vertext to predict -- start/end)
size - numeric value (number of rows)
"""
dist = RandomDistribution(seed=seed)
err_halfnorm = dist.return_half_normal_variables(location=0, scale=variance_halfnorm**0.5, size=size)
dist = RandomDistribution(seed=seed)
err_norm = dist.return_normal_variables(location=0, scale=variance_norm**0.5, size=size)
if vertex == 'start':
return err_norm + err_halfnorm
if vertex == 'end':
return err_norm - err_halfnorm
def testvalues(self):
model = LinearRegressionModel(self.specification, self.errorspecification)
pred_value = model.calc_predvalue(self.data)
expected_act = self.data.calculate_equation(self.specification.coefficients[0])
expected_act.shape = (4,1)
dist = RandomDistribution(seed=1)
pred_act = dist.return_normal_variables(location=expected_act, scale=1, size=(4,1))
pred_diff = all(pred_value.data == pred_act)
self.assertEqual(True, pred_diff)
def calc_errorcomponent(self, size, mean=0, sd=1, seed=1):
"""
The method returns the contribution of the error in the calculation
of the predicted value for the different choices.
Inputs:
size - numeric value (number of rows)
mean - numeric value (mean)
sd - numeric value (standard deviation)
"""
dist = RandomDistribution(seed=seed)
err_norm = dist.return_normal_variables(location=mean, scale=sd, size=size)
return err_norm
def testvalues(self):
model = StocFronRegressionModel(self.specification, self.errorspecification)
pred_value = model.calc_predvalue(self.data)
expected_act = self.data.calculate_equation(self.specification.coefficients[0])
expected_act.shape = (4, 1)
variance = self.errorspecification.variance
dist = RandomDistribution(seed=1)
err_norm = dist.return_normal_variables(location=0, scale=variance[0, 0] ** 0.5, size=(4, 1))
pred_act = expected_act + err_norm
pred_diff = all(pred_value.data == pred_act)
self.assertEquals(True, pred_diff)
def calc_halfnormal_error(self, threshold, limit, seed, size): """ A draw from a half normal distribution with 3 s.d. = abs(threshold - limit) For smoothing about the boundaries """ dist = RandomDistribution(seed=seed) assu_scale = abs(threshold-limit)/3.0 err_halfnorm = dist.return_half_normal_variables(location=0, scale=assu_scale, size=size) chkRowsMore = err_halfnorm > abs(threshold-limit) err_halfnorm[chkRowsMore] = abs(threshold-limit) return err_halfnorm
def generate_random_numbers(self):
"""
The method generates a random array for making the choice of
the alternative.
Inputs:
None
"""
#random.seed(seed=self.seed)
#err = random.random((3,1))
#f = open('test_res', 'a')
#f.write('probability - %s' %self.seed)
#f.write(str(list(err[:3,:])))
#f.write('\n')
#f.close()
dist = RandomDistribution(self.seed)
rand_numbers = dist.return_random_variables(self.num_agents)
return rand_numbers
def testvalues(self):
model = StocFronRegressionModel(self.specification,
self.errorspecification)
pred_value = model.calc_predvalue(self.data)
expected_act = self.data.calculate_equation(
self.specification.coefficients[0])
expected_act.shape = (4, 1)
variance = self.errorspecification.variance
dist = RandomDistribution(seed=1)
err_norm = dist.return_normal_variables(location=0,
scale=variance[0, 0]**0.5,
size=(4, 1))
pred_act = (expected_act + err_norm)
pred_diff = all(pred_value.data == pred_act)
self.assertEquals(True, pred_diff)
def generate_random_numbers(self):
"""
The method generates a random array for making the choice of
the alternative.
Inputs:
None
"""
#random.seed(seed=self.seed)
#err = random.random((3,1))
#f = open('test_res', 'a')
#f.write('probability - %s' %self.seed)
#f.write(str(list(err[:3,:])))
#f.write('\n')
#f.close()
dist = RandomDistribution(self.seed)
rand_numbers = dist.return_random_variables(self.num_agents)
return rand_numbers
def calc_errorcomponent(self, variance_norm, variance_halfnorm, vertex,
size, seed):
"""
The method returns the contribution of the error component in the
calculation of the predicted value for the different choices.
Inputs:
variance_norm - numeric value (variance of the normal portion of error)
variance_halfnorm - numeric value (variance of the half normal portion of
error)
vertex - string (the vertext to predict -- start/end)
size - numeric value (number of rows)
"""
dist = RandomDistribution(seed=seed)
err_halfnorm = dist.return_half_normal_variables(
location=0, scale=variance_halfnorm**0.5, size=size)
dist = RandomDistribution(seed=seed)
err_norm = dist.return_normal_variables(location=0,
scale=variance_norm**0.5,
size=size)
if vertex == 'start':
return err_norm + err_halfnorm
if vertex == 'end':
return err_norm - err_halfnorm
