From 9c88536f3488bc2c7f07bffb66659f7892447cc3 Mon Sep 17 00:00:00 2001
From: Le Roux Erwan <erwan.le-roux@irstea.fr>
Date: Mon, 23 Mar 2020 21:47:48 +0100
Subject: [PATCH] [quantile regression project] add annual maxima osbervation
from daily exponential. add sammple_r_function as argument to
rmargin_from_nb_obs. add abstract extreme params
---
.../distribution/abstract_extreme_params.py | 20 +++++++++++++
extreme_fit/distribution/abstract_params.py | 13 ++-------
extreme_fit/distribution/exp_params.py | 17 ++++++++---
extreme_fit/distribution/gev/gev_params.py | 3 +-
extreme_fit/distribution/gpd/gpd_params.py | 3 +-
.../abstract_margin_function.py | 3 +-
.../combined_margin_function.py | 2 +-
.../independent_margin_function.py | 14 ++++-----
.../margin_function/linear_margin_function.py | 6 ++--
.../parametric_margin_function.py | 7 +++--
.../margin_model/abstract_margin_model.py | 8 +++--
.../abstract_temporal_linear_margin_model.py | 6 ++--
.../linear_margin_model.py | 29 ++++++++++---------
.../margin_model/parametric_margin_model.py | 5 ++--
.../daily_observations.py | 3 +-
.../test_spatio_temporal_observations.py | 15 +++++++---
16 files changed, 98 insertions(+), 56 deletions(-)
create mode 100644 extreme_fit/distribution/abstract_extreme_params.py
diff --git a/extreme_fit/distribution/abstract_extreme_params.py b/extreme_fit/distribution/abstract_extreme_params.py
new file mode 100644
index 00000000..53cd1147
--- /dev/null
+++ b/extreme_fit/distribution/abstract_extreme_params.py
@@ -0,0 +1,20 @@
+from extreme_fit.distribution.abstract_params import AbstractParams
+
+
+class AbstractExtremeParams(AbstractParams):
+ pass
+
+ # Extreme parameters
+ SCALE = 'scale'
+ LOC = 'loc'
+ SHAPE = 'shape'
+
+ def __init__(self, loc: float, scale: float, shape: float):
+ self.location = loc
+ self.scale = scale
+ self.shape = shape
+ # By default, scale cannot be negative
+ # (sometimes it happens, when we want to find a quantile for every point of a 2D map
+ # then it can happen that a corner point that was not used for fitting correspond to a negative scale,
+ # in the case we set all the parameters as equal to np.nan, and we will not display those points)
+ self.has_undefined_parameters = self.scale <= 0
\ No newline at end of file
diff --git a/extreme_fit/distribution/abstract_params.py b/extreme_fit/distribution/abstract_params.py
index f367a820..631f8656 100644
--- a/extreme_fit/distribution/abstract_params.py
+++ b/extreme_fit/distribution/abstract_params.py
@@ -17,21 +17,14 @@ class AbstractParams(object):
QUANTILE_COLORS = ['orange', 'red', 'darkviolet']
# Summary
SUMMARY_NAMES = PARAM_NAMES + QUANTILE_NAMES
+
+ # Simple parameters
+ RATE = 'rate'
# Extreme parameters
SCALE = 'scale'
LOC = 'loc'
SHAPE = 'shape'
- def __init__(self, loc: float, scale: float, shape: float):
- self.location = loc
- self.scale = scale
- self.shape = shape
- # By default, scale cannot be negative
- # (sometimes it happens, when we want to find a quantile for every point of a 2D map
- # then it can happen that a corner point that was not used for fitting correspond to a negative scale,
- # in the case we set all the parameters as equal to np.nan, and we will not display those points)
- self.has_undefined_parameters = self.scale <= 0
-
@classmethod
def from_dict(cls, params: dict):
return cls(**params)
diff --git a/extreme_fit/distribution/exp_params.py b/extreme_fit/distribution/exp_params.py
index 6a003ee2..5ffb82f9 100644
--- a/extreme_fit/distribution/exp_params.py
+++ b/extreme_fit/distribution/exp_params.py
@@ -1,9 +1,18 @@
+import numpy as np
+
from extreme_fit.distribution.abstract_params import AbstractParams
-class ExpParams(object):
- PARAM_NAMES = [AbstractParams.SCALE]
+class ExpParams(AbstractParams):
+ PARAM_NAMES = [AbstractParams.RATE]
- def __init__(self, scale: float):
- self.scale = scale
+ def __init__(self, rate) -> None:
+ self.rate = rate
+ self.has_undefined_parameters = self.rate < 0
+ @property
+ def param_values(self):
+ if self.has_undefined_parameters:
+ return [np.nan for _ in range(1)]
+ else:
+ return [self.rate]
diff --git a/extreme_fit/distribution/gev/gev_params.py b/extreme_fit/distribution/gev/gev_params.py
index db84b878..a116f245 100644
--- a/extreme_fit/distribution/gev/gev_params.py
+++ b/extreme_fit/distribution/gev/gev_params.py
@@ -5,13 +5,14 @@ from typing import List
from cached_property import cached_property
from mpmath import euler, pi
+from extreme_fit.distribution.abstract_extreme_params import AbstractExtremeParams
from extreme_fit.distribution.abstract_params import AbstractParams
from extreme_fit.model.utils import r
import numpy as np
from scipy.special import gamma
-class GevParams(AbstractParams):
+class GevParams(AbstractExtremeParams):
# Parameters
PARAM_NAMES = [AbstractParams.LOC, AbstractParams.SCALE, AbstractParams.SHAPE]
# Summary
diff --git a/extreme_fit/distribution/gpd/gpd_params.py b/extreme_fit/distribution/gpd/gpd_params.py
index c440415f..ed6b1ac9 100644
--- a/extreme_fit/distribution/gpd/gpd_params.py
+++ b/extreme_fit/distribution/gpd/gpd_params.py
@@ -1,8 +1,9 @@
+from extreme_fit.distribution.abstract_extreme_params import AbstractExtremeParams
from extreme_fit.distribution.abstract_params import AbstractParams
from extreme_fit.model.utils import r
-class GpdParams(AbstractParams):
+class GpdParams(AbstractExtremeParams):
# TODO: understand better why the gpdfit return 2 parameters, alors que d'un autre cote d autres definitions de la distribution parlent d un parametre location
# Parameters
diff --git a/extreme_fit/function/margin_function/abstract_margin_function.py b/extreme_fit/function/margin_function/abstract_margin_function.py
index df30892b..860d25c5 100644
--- a/extreme_fit/function/margin_function/abstract_margin_function.py
+++ b/extreme_fit/function/margin_function/abstract_margin_function.py
@@ -20,8 +20,9 @@ class AbstractMarginFunction(AbstractFunction):
VISUALIZATION_RESOLUTION = 100
VISUALIZATION_TEMPORAL_STEPS = 2
- def __init__(self, coordinates: AbstractCoordinates):
+ def __init__(self, coordinates: AbstractCoordinates, params_class: type = GevParams):
super().__init__(coordinates)
+ self.params_class = params_class
self.mask_2D = None
# Visualization parameters
diff --git a/extreme_fit/function/margin_function/combined_margin_function.py b/extreme_fit/function/margin_function/combined_margin_function.py
index 08737bc1..20b5e474 100644
--- a/extreme_fit/function/margin_function/combined_margin_function.py
+++ b/extreme_fit/function/margin_function/combined_margin_function.py
@@ -18,7 +18,7 @@ class CombinedMarginFunction(AbstractMarginFunction):
def get_gev_params(self, coordinate: np.ndarray) -> GevParams:
gev_params_list = [margin_function.get_gev_params(coordinate) for margin_function in self.margin_functions]
mean_gev_params = np.mean(np.array([gev_param.to_array() for gev_param in gev_params_list]), axis=0)
- gev_param = GevParams(*mean_gev_params)
+ gev_param = self.params_class(*mean_gev_params)
return gev_param
@classmethod
diff --git a/extreme_fit/function/margin_function/independent_margin_function.py b/extreme_fit/function/margin_function/independent_margin_function.py
index ccb6dba8..43169d71 100644
--- a/extreme_fit/function/margin_function/independent_margin_function.py
+++ b/extreme_fit/function/margin_function/independent_margin_function.py
@@ -14,9 +14,9 @@ class IndependentMarginFunction(AbstractMarginFunction):
IndependentMarginFunction: each parameter of the GEV are modeled independently
"""
- def __init__(self, coordinates: AbstractCoordinates):
+ def __init__(self, coordinates: AbstractCoordinates, params_class: type = GevParams):
"""Attribute 'gev_param_name_to_param_function' maps each GEV parameter to its corresponding function"""
- super().__init__(coordinates)
+ super().__init__(coordinates, params_class)
self.gev_param_name_to_param_function = None # type: Union[None, Dict[str, AbstractParamFunction]]
def get_gev_params(self, coordinate: np.ndarray, is_transformed: bool = True) -> GevParams:
@@ -24,13 +24,11 @@ class IndependentMarginFunction(AbstractMarginFunction):
# Since all the coordinates are usually transformed by default
# then we assume that the input coordinate are transformed by default
assert self.gev_param_name_to_param_function is not None
- assert len(self.gev_param_name_to_param_function) == 3
+ assert len(self.gev_param_name_to_param_function) == len(self.params_class.PARAM_NAMES)
transformed_coordinate = coordinate if is_transformed else self.transform(coordinate)
- gev_params = {}
- for gev_param_name in GevParams.PARAM_NAMES:
- param_function = self.gev_param_name_to_param_function[gev_param_name]
- gev_params[gev_param_name] = param_function.get_param_value(transformed_coordinate)
- return GevParams.from_dict(gev_params)
+ params = {param_name: param_function.get_param_value(transformed_coordinate)
+ for param_name, param_function in self.gev_param_name_to_param_function.items()}
+ return self.params_class.from_dict(params)
diff --git a/extreme_fit/function/margin_function/linear_margin_function.py b/extreme_fit/function/margin_function/linear_margin_function.py
index d89a7a59..3cdc1f84 100644
--- a/extreme_fit/function/margin_function/linear_margin_function.py
+++ b/extreme_fit/function/margin_function/linear_margin_function.py
@@ -28,9 +28,11 @@ class LinearMarginFunction(ParametricMarginFunction):
COEF_CLASS = LinearCoef
def __init__(self, coordinates: AbstractCoordinates, gev_param_name_to_dims: Dict[str, List[int]],
- gev_param_name_to_coef: Dict[str, AbstractCoef], starting_point: Union[None, int] = None):
+ gev_param_name_to_coef: Dict[str, AbstractCoef], starting_point: Union[None, int] = None,
+ params_class: type = GevParams):
self.gev_param_name_to_coef = None # type: Union[None, Dict[str, LinearCoef]]
- super().__init__(coordinates, gev_param_name_to_dims, gev_param_name_to_coef, starting_point)
+ super().__init__(coordinates, gev_param_name_to_dims, gev_param_name_to_coef, starting_point,
+ params_class)
def load_specific_param_function(self, gev_param_name) -> AbstractParamFunction:
return LinearParamFunction(dims=self.gev_param_name_to_dims[gev_param_name],
diff --git a/extreme_fit/function/margin_function/parametric_margin_function.py b/extreme_fit/function/margin_function/parametric_margin_function.py
index 57037216..60a1bbb8 100644
--- a/extreme_fit/function/margin_function/parametric_margin_function.py
+++ b/extreme_fit/function/margin_function/parametric_margin_function.py
@@ -31,10 +31,11 @@ class ParametricMarginFunction(IndependentMarginFunction):
COEF_CLASS = None
def __init__(self, coordinates: AbstractCoordinates, gev_param_name_to_dims: Dict[str, List[int]],
- gev_param_name_to_coef: Dict[str, AbstractCoef], starting_point: Union[None, int] = None):
+ gev_param_name_to_coef: Dict[str, AbstractCoef], starting_point: Union[None, int] = None,
+ params_class: type = GevParams):
# Starting point for the trend is the same for all the parameters
self.starting_point = starting_point
- super().__init__(coordinates)
+ super().__init__(coordinates, params_class)
self.gev_param_name_to_dims = gev_param_name_to_dims # type: Dict[str, List[int]]
# Check the dimension are well-defined with respect to the coordinates
@@ -48,7 +49,7 @@ class ParametricMarginFunction(IndependentMarginFunction):
# Build gev_parameter_to_param_function dictionary
self.gev_param_name_to_param_function = {} # type: Dict[str, AbstractParamFunction]
# Map each gev_param_name to its corresponding param_function
- for gev_param_name in GevParams.PARAM_NAMES:
+ for gev_param_name in self.params_class.PARAM_NAMES:
# By default, if dims are not specified, a constantParamFunction is chosen
if self.gev_param_name_to_dims.get(gev_param_name) is None:
param_function = ConstantParamFunction(constant=self.gev_param_name_to_coef[gev_param_name].intercept)
diff --git a/extreme_fit/model/margin_model/abstract_margin_model.py b/extreme_fit/model/margin_model/abstract_margin_model.py
index b4ac6918..79638873 100644
--- a/extreme_fit/model/margin_model/abstract_margin_model.py
+++ b/extreme_fit/model/margin_model/abstract_margin_model.py
@@ -20,12 +20,14 @@ class AbstractMarginModel(AbstractModel, ABC):
"""
def __init__(self, coordinates: AbstractCoordinates, use_start_value=False,
- params_start_fit=None, params_sample=None):
+ params_start_fit=None, params_sample=None,
+ params_class=GevParams):
super().__init__(use_start_value, params_start_fit, params_sample)
assert isinstance(coordinates, AbstractCoordinates), type(coordinates)
self.coordinates = coordinates
self.margin_function_sample = None # type: AbstractMarginFunction
self.margin_function_start_fit = None # type: AbstractMarginFunction
+ self.params_class = params_class
self.load_margin_functions()
def load_margin_functions(self):
@@ -34,9 +36,9 @@ class AbstractMarginModel(AbstractModel, ABC):
def default_load_margin_functions(self, margin_function_class):
# todo: check it i could remove these attributes
self.margin_function_sample = margin_function_class(coordinates=self.coordinates,
- default_params=GevParams.from_dict(self.params_sample))
+ default_params=self.params_class.from_dict(self.params_sample))
self.margin_function_start_fit = margin_function_class(coordinates=self.coordinates,
- default_params=GevParams.from_dict(
+ default_params=self.params_class.from_dict(
self.params_start_fit))
# Conversion class methods
diff --git a/extreme_fit/model/margin_model/linear_margin_model/abstract_temporal_linear_margin_model.py b/extreme_fit/model/margin_model/linear_margin_model/abstract_temporal_linear_margin_model.py
index e5e2ee45..69f6531e 100644
--- a/extreme_fit/model/margin_model/linear_margin_model/abstract_temporal_linear_margin_model.py
+++ b/extreme_fit/model/margin_model/linear_margin_model/abstract_temporal_linear_margin_model.py
@@ -29,8 +29,10 @@ class AbstractTemporalLinearMarginModel(LinearMarginModel):
fit_method=TemporalMarginFitMethod.is_mev_gev_fit,
nb_iterations_for_bayesian_fit=5000,
params_start_fit_bayesian=None,
- type_for_MLE="GEV"):
- super().__init__(coordinates, use_start_value, params_start_fit, params_sample, starting_point)
+ type_for_MLE="GEV",
+ params_class=GevParams):
+ super().__init__(coordinates, use_start_value, params_start_fit, params_sample, starting_point,
+ params_class)
self.type_for_mle = type_for_MLE
self.params_start_fit_bayesian = params_start_fit_bayesian
self.nb_iterations_for_bayesian_fit = nb_iterations_for_bayesian_fit
diff --git a/extreme_fit/model/margin_model/linear_margin_model/linear_margin_model.py b/extreme_fit/model/margin_model/linear_margin_model/linear_margin_model.py
index 08ac9900..422bb806 100644
--- a/extreme_fit/model/margin_model/linear_margin_model/linear_margin_model.py
+++ b/extreme_fit/model/margin_model/linear_margin_model/linear_margin_model.py
@@ -7,12 +7,12 @@ from extreme_fit.distribution.gev.gev_params import GevParams
class LinearMarginModel(ParametricMarginModel):
@classmethod
- def from_coef_list(cls, coordinates, gev_param_name_to_coef_list, **kwargs):
+ def from_coef_list(cls, coordinates, gev_param_name_to_coef_list, params_class=GevParams, **kwargs):
params = {}
for param_name, coef_list in gev_param_name_to_coef_list.items():
for idx, coef in enumerate(coef_list, -1):
params[(param_name, idx)] = coef
- return cls(coordinates, params_sample=params, params_start_fit=params, **kwargs)
+ return cls(coordinates, params_sample=params, params_start_fit=params, params_class=params_class, **kwargs)
def load_margin_functions(self, gev_param_name_to_dims=None):
assert gev_param_name_to_dims is not None, 'LinearMarginModel cannot be used for sampling/fitting \n' \
@@ -27,34 +27,37 @@ class LinearMarginModel(ParametricMarginModel):
self.margin_function_sample = LinearMarginFunction(coordinates=self.coordinates,
gev_param_name_to_coef=coef_sample,
gev_param_name_to_dims=gev_param_name_to_dims,
- starting_point=self.starting_point)
+ starting_point=self.starting_point,
+ params_class=self.params_class)
# Load start fit coef
coef_start_fit = self.gev_param_name_to_linear_coef(param_name_and_dim_to_coef=self.params_start_fit)
self.margin_function_start_fit = LinearMarginFunction(coordinates=self.coordinates,
gev_param_name_to_coef=coef_start_fit,
gev_param_name_to_dims=gev_param_name_to_dims,
- starting_point=self.starting_point)
+ starting_point=self.starting_point,
+ params_class=self.params_class)
@property
def default_param_name_and_dim_to_coef(self) -> dict:
default_intercept = 1
default_slope = 0.01
gev_param_name_and_dim_to_coef = {}
- for gev_param_name in GevParams.PARAM_NAMES:
- gev_param_name_and_dim_to_coef[(gev_param_name, -1)] = default_intercept
+ for param_name in self.params_class.PARAM_NAMES:
+ gev_param_name_and_dim_to_coef[(param_name, -1)] = default_intercept
for dim in self.coordinates.coordinates_dims:
- gev_param_name_and_dim_to_coef[(gev_param_name, dim)] = default_slope
+ gev_param_name_and_dim_to_coef[(param_name, dim)] = default_slope
return gev_param_name_and_dim_to_coef
def gev_param_name_to_linear_coef(self, param_name_and_dim_to_coef):
- gev_param_name_to_linear_coef = {}
- for gev_param_name in GevParams.PARAM_NAMES:
- idx_to_coef = {idx: param_name_and_dim_to_coef[(gev_param_name, idx)] for idx in
+ param_name_to_linear_coef = {}
+ param_names = list(set([e[0] for e in param_name_and_dim_to_coef.keys()]))
+ for param_name in param_names:
+ idx_to_coef = {idx: param_name_and_dim_to_coef[(param_name, idx)] for idx in
[-1] + self.coordinates.coordinates_dims}
- linear_coef = LinearCoef(gev_param_name=gev_param_name, idx_to_coef=idx_to_coef)
- gev_param_name_to_linear_coef[gev_param_name] = linear_coef
- return gev_param_name_to_linear_coef
+ linear_coef = LinearCoef(gev_param_name=param_name, idx_to_coef=idx_to_coef)
+ param_name_to_linear_coef[param_name] = linear_coef
+ return param_name_to_linear_coef
class ConstantMarginModel(LinearMarginModel):
diff --git a/extreme_fit/model/margin_model/parametric_margin_model.py b/extreme_fit/model/margin_model/parametric_margin_model.py
index 61279f4a..724366c4 100644
--- a/extreme_fit/model/margin_model/parametric_margin_model.py
+++ b/extreme_fit/model/margin_model/parametric_margin_model.py
@@ -3,6 +3,7 @@ from abc import ABC
import numpy as np
import pandas as pd
+from extreme_fit.distribution.gev.gev_params import GevParams
from extreme_fit.function.margin_function.parametric_margin_function import \
ParametricMarginFunction
from extreme_fit.model.result_from_model_fit.result_from_spatial_extreme import ResultFromSpatialExtreme
@@ -15,14 +16,14 @@ from spatio_temporal_dataset.coordinates.abstract_coordinates import AbstractCoo
class ParametricMarginModel(AbstractMarginModel, ABC):
def __init__(self, coordinates: AbstractCoordinates, use_start_value=False, params_start_fit=None,
- params_sample=None, starting_point=None):
+ params_sample=None, starting_point=None, params_class=GevParams):
"""
:param starting_point: starting coordinate for the temporal trend
"""
self.starting_point = starting_point
self.margin_function_sample = None # type: ParametricMarginFunction
self.margin_function_start_fit = None # type: ParametricMarginFunction
- super().__init__(coordinates, use_start_value, params_start_fit, params_sample)
+ super().__init__(coordinates, use_start_value, params_start_fit, params_sample, params_class)
def fitmargin_from_maxima_gev(self, data: np.ndarray, df_coordinates_spat: pd.DataFrame,
df_coordinates_temp: pd.DataFrame) -> ResultFromSpatialExtreme:
diff --git a/spatio_temporal_dataset/spatio_temporal_observations/daily_observations.py b/spatio_temporal_dataset/spatio_temporal_observations/daily_observations.py
index 9483f3b3..b0044f33 100644
--- a/spatio_temporal_dataset/spatio_temporal_observations/daily_observations.py
+++ b/spatio_temporal_dataset/spatio_temporal_observations/daily_observations.py
@@ -21,6 +21,7 @@ class DailyExp(AbstractSpatioTemporalObservations):
def from_sampling(cls, nb_obs: int, coordinates: AbstractCoordinates,
margin_model: AbstractMarginModel):
exponential_values = margin_model.rmargin_from_nb_obs(nb_obs=nb_obs,
- coordinates_values=coordinates.coordinates_values())
+ coordinates_values=coordinates.coordinates_values(),
+ sample_r_function='rexp')
df_exponential_values = pd.DataFrame(data=exponential_values, index=coordinates.index)
return cls(df_maxima_gev=df_exponential_values)
diff --git a/test/test_spatio_temporal_dataset/test_spatio_temporal_observations.py b/test/test_spatio_temporal_dataset/test_spatio_temporal_observations.py
index 716a5856..4786e54c 100644
--- a/test/test_spatio_temporal_dataset/test_spatio_temporal_observations.py
+++ b/test/test_spatio_temporal_dataset/test_spatio_temporal_observations.py
@@ -4,6 +4,7 @@ import numpy as np
import pandas as pd
from extreme_fit.distribution.abstract_params import AbstractParams
+from extreme_fit.distribution.exp_params import ExpParams
from extreme_fit.model.margin_model.linear_margin_model.temporal_linear_margin_models import StationaryTemporalModel
from extreme_fit.model.utils import set_seed_for_test
from spatio_temporal_dataset.coordinates.temporal_coordinates.generated_temporal_coordinates import \
@@ -33,18 +34,24 @@ class TestDailyObservations(unittest.TestCase):
set_seed_for_test(seed=42)
self.coordinates = ConsecutiveTemporalCoordinates.from_nb_temporal_steps(nb_temporal_steps=10)
gev_param_name_to_coef_list = {
- AbstractParams.SCALE: [1],
+ AbstractParams.RATE: [1],
}
- self.margin_model = StationaryTemporalModel.from_coef_list(self.coordinates, gev_param_name_to_coef_list)
+ self.margin_model = StationaryTemporalModel.from_coef_list(self.coordinates, gev_param_name_to_coef_list,
+ params_class=ExpParams)
+
+ def test_instance_exp_params(self):
+ last_coordinate = self.coordinates.coordinates_values()[-1]
+ params = self.margin_model.margin_function_sample.get_gev_params(last_coordinate)
+ self.assertIsInstance(params, ExpParams)
def test_exponential_observations(self):
obs = DailyExp.from_sampling(nb_obs=1, coordinates=self.coordinates,
margin_model=self.margin_model)
- self.assertAlmostEqual(obs.df_maxima_gev.mean()[0], 4.692385276235156)
+ self.assertAlmostEqual(obs.df_maxima_gev.mean()[0], 0.574829320536985)
def test_annual_maxima_observations_from_daily_observations(self):
obs = DailyExpAnnualMaxima.from_sampling(1, self.coordinates, self.margin_model)
- self.assertAlmostEqual(obs.df_maxima_gev.mean()[0], 1183.8468374768636)
+ self.assertAlmostEqual(obs.df_maxima_gev.mean()[0], 6.523848726794694)
if __name__ == '__main__':
--
GitLab