def _add_bounds_to_percentiles_and_forecast_at_percentiles(
self, percentiles, forecast_at_percentiles, bounds_pairing):
"""
Padding of the lower and upper bounds of the percentiles for a
given phenomenon, and padding of forecast values using the
constant lower and upper bounds.
Args:
percentiles (numpy.ndarray):
Array of percentiles from a Cumulative Distribution Function.
forecast_at_percentiles (numpy.ndarray):
Array containing the underlying forecast values at each
percentile.
bounds_pairing (tuple):
Lower and upper bound to be used as the ends of the
cumulative distribution function.
Raises:
ValueError: If the percentile points are outside the ECC bounds
and self.ecc_bounds_warning is False.
ValueError: If the percentiles are not in ascending order.
Warns:
Warning: If the percentile points are outside the ECC bounds
and self.ecc_bounds_warning is True.
"""
lower_bound, upper_bound = bounds_pairing
percentiles = insert_lower_and_upper_endpoint_to_1d_array(
percentiles, 0, 100)
forecast_at_percentiles_with_endpoints = \
concatenate_2d_array_with_2d_array_endpoints(
forecast_at_percentiles, lower_bound, upper_bound)
if np.any(np.diff(forecast_at_percentiles_with_endpoints) < 0):
msg = ("The end points added to the forecast at percentile "
"values representing each percentile must result in "
"an ascending order. "
"In this case, the forecast at percentile values {} "
"is outside the allowable range given by the "
"bounds {}".format(forecast_at_percentiles, bounds_pairing))
if self.ecc_bounds_warning:
warn_msg = msg + (" The percentile values that have "
"exceeded the existing bounds will be used "
"as new bounds.")
warnings.warn(warn_msg)
if upper_bound < forecast_at_percentiles_with_endpoints.max():
upper_bound = forecast_at_percentiles_with_endpoints.max()
if lower_bound > forecast_at_percentiles_with_endpoints.min():
lower_bound = forecast_at_percentiles_with_endpoints.min()
forecast_at_percentiles_with_endpoints = \
concatenate_2d_array_with_2d_array_endpoints(
forecast_at_percentiles, lower_bound, upper_bound)
else:
raise ValueError(msg)
if np.any(np.diff(percentiles) < 0):
msg = ("The percentiles must be in ascending order."
"The input percentiles were {}".format(percentiles))
raise ValueError(msg)
return percentiles, forecast_at_percentiles_with_endpoints
def test_3d_input(self):
"""
Test that a 3d input array results in the expected error.
"""
input_array = np.array([[[-40, 200, 1000]]])
msg = "all the input arrays must have same number of dimensions"
with self.assertRaisesRegex(ValueError, msg):
concatenate_2d_array_with_2d_array_endpoints(
input_array, -100, 10000)
def _add_bounds_to_thresholds_and_probabilities(self, threshold_points,
probabilities_for_cdf,
bounds_pairing):
"""
Padding of the lower and upper bounds of the distribution for a
given phenomenon for the threshold_points, and padding of
probabilities of 0 and 1 to the forecast probabilities.
Args:
threshold_points (Numpy array):
Array of threshold values used to calculate the probabilities.
probabilities_for_cdf (Numpy array):
Array containing the probabilities used for constructing an
cumulative distribution function i.e. probabilities
below threshold.
bounds_pairing (Tuple):
Lower and upper bound to be used as the ends of the
cumulative distribution function.
Returns:
(tuple) : tuple containing:
**threshold_points** (Numpy array):
Array of threshold values padded with the lower and upper
bound of the distribution.
**probabilities_for_cdf** (Numpy array):
Array containing the probabilities padded with 0 and 1 at
each end.
"""
lower_bound, upper_bound = bounds_pairing
threshold_points_with_endpoints = \
insert_lower_and_upper_endpoint_to_1d_array(
threshold_points, lower_bound, upper_bound)
probabilities_for_cdf = concatenate_2d_array_with_2d_array_endpoints(
probabilities_for_cdf, 0, 1)
if np.any(np.diff(threshold_points_with_endpoints) < 0):
msg = ("The calculated threshold values {} are not in ascending "
"order as required for the cumulative distribution "
"function (CDF). This is due to the threshold values "
"exceeding the range given by the ECC bounds {}.".format(
threshold_points_with_endpoints, bounds_pairing))
# If ecc_bounds_warning has been set, generate a warning message
# rather than raising an exception so that subsequent processing
# can continue. Then apply the new bounds as necessary to
# ensure the threshold values and endpoints are in ascending
# order and avoid problems further along the processing chain.
if self.ecc_bounds_warning:
warn_msg = msg + (" The threshold points that have "
"exceeded the existing bounds will be used "
"as new bounds.")
warnings.warn(warn_msg)
if upper_bound < max(threshold_points_with_endpoints):
upper_bound = max(threshold_points_with_endpoints)
if lower_bound > min(threshold_points_with_endpoints):
lower_bound = min(threshold_points_with_endpoints)
threshold_points_with_endpoints = \
insert_lower_and_upper_endpoint_to_1d_array(
threshold_points, lower_bound, upper_bound)
else:
raise ValueError(msg)
return threshold_points_with_endpoints, probabilities_for_cdf
def test_basic(self):
"""
Basic test that the result is a numpy array with the expected contents.
"""
expected = np.array([[0, 20, 50, 80, 100]])
input_array = np.array([[20, 50, 80]])
result = concatenate_2d_array_with_2d_array_endpoints(
input_array, 0, 100)
self.assertIsInstance(result, np.ndarray)
self.assertArrayAlmostEqual(result, expected)
def test_another_example(self):
"""
Another basic test that the result is a numpy array with the
expected contents.
"""
expected = np.array(
[[-100, -40, 200, 1000, 10000], [-100, -40, 200, 1000, 10000]])
input_array = np.array([[-40, 200, 1000], [-40, 200, 1000]])
result = concatenate_2d_array_with_2d_array_endpoints(
input_array, -100, 10000)
self.assertIsInstance(result, np.ndarray)
self.assertArrayAlmostEqual(result, expected)
def _add_bounds_to_percentiles_and_forecast_at_percentiles(
percentiles, forecast_at_percentiles, bounds_pairing):
"""
Padding of the lower and upper bounds of the percentiles for a
given phenomenon, and padding of forecast values using the
constant lower and upper bounds.
Args:
percentiles (Numpy array):
Array of percentiles from a Cumulative Distribution Function.
forecast_at_percentiles (Numpy array):
Array containing the underlying forecast values at each
percentile.
bounds_pairing (Tuple):
Lower and upper bound to be used as the ends of the
cumulative distribution function.
Returns:
(tuple) : tuple containing:
**percentiles** (Numpy array):
Array of percentiles from a Cumulative Distribution
Function.
**forecast_at_percentiles** (Numpy array):
Array containing the underlying forecast values at each
percentile.
"""
lower_bound, upper_bound = bounds_pairing
percentiles = insert_lower_and_upper_endpoint_to_1d_array(
percentiles, 0, 100)
forecast_at_percentiles = concatenate_2d_array_with_2d_array_endpoints(
forecast_at_percentiles, lower_bound, upper_bound)
if np.any(np.diff(forecast_at_percentiles) < 0):
msg = ("The end points added to the forecast at percentiles "
"values representing each percentile must result in "
"an ascending order. "
"In this case, the forecast at percentile values {} "
"is outside the allowable range given by the "
"bounds {}".format(
forecast_at_percentiles, bounds_pairing))
raise ValueError(msg)
if np.any(np.diff(percentiles) < 0):
msg = ("The percentiles must be in ascending order."
"The input percentiles were {}".format(percentiles))
raise ValueError(msg)
return percentiles, forecast_at_percentiles
def _add_bounds_to_thresholds_and_probabilities(
threshold_points, probabilities_for_cdf, bounds_pairing):
"""
Padding of the lower and upper bounds of the distribution for a
given phenomenon for the threshold_points, and padding of
probabilities of 0 and 1 to the forecast probabilities.
Args:
threshold_points (Numpy array):
Array of threshold values used to calculate the probabilities.
probabilities_for_cdf (Numpy array):
Array containing the probabilities used for constructing an
cumulative distribution function i.e. probabilities
below threshold.
bounds_pairing (Tuple):
Lower and upper bound to be used as the ends of the
cumulative distribution function.
Returns:
(tuple) : tuple containing:
**threshold_points** (Numpy array):
Array of threshold values padded with the lower and upper
bound of the distribution.
**probabilities_for_cdf** (Numpy array):
Array containing the probabilities padded with 0 and 1 at
each end.
"""
lower_bound, upper_bound = bounds_pairing
threshold_points = insert_lower_and_upper_endpoint_to_1d_array(
threshold_points, lower_bound, upper_bound)
probabilities_for_cdf = concatenate_2d_array_with_2d_array_endpoints(
probabilities_for_cdf, 0, 1)
if np.any(np.diff(threshold_points) < 0):
msg = ("The end points added to the threshold values for "
"constructing the Cumulative Distribution Function (CDF) "
"must result in an ascending order. "
"In this case, the threshold points {} must be "
"outside the allowable range given by the "
"bounds {}".format(
threshold_points, bounds_pairing))
raise ValueError(msg)
return threshold_points, probabilities_for_cdf
