"""Created on Aug 10 23:08:38 2024"""
import itertools
import warnings
from typing import Optional, Tuple, Union, List, Callable, Any, Sequence
import matplotlib.pyplot as plt
import numpy as np
from matplotlib.axes import Axes
from mpyez.backend.uPlotting import LinePlot # type: ignore
from mpyez.ezPlotting import plot_xy # type: ignore
from scipy.optimize import Bounds, curve_fit
# importing from files to avoid circular import
from .chiSquare_f import ChiSquareFitter
from .exponential_f import ExponentialFitter
from .foldedNormal_f import FoldedNormalFitter
from .gamma_f import GammaFitter
from .gaussian_f import GaussianFitter
from .halfNormal_f import HalfNormalFitter
from .laplace_f import LaplaceFitter
from .logNormal_f import LogNormalFitter
from .polynomial_f import LineFitter
from .skewNormal_f import SkewNormalFitter
from .utilities_f import sanity_check, _plot_fit
from .. import (
epsilon,
GAUSSIAN,
LAPLACE,
LINE,
LOG_NORMAL,
SKEW_NORMAL,
CHI_SQUARE,
EXPONENTIAL,
FOLDED_NORMAL,
GAMMA,
NORMAL,
HALF_NORMAL,
OneDArray,
Params_,
)
# mock initialize the internal classes for auto MixedDataFitter class
fitter_dict = {
CHI_SQUARE: ChiSquareFitter,
EXPONENTIAL: ExponentialFitter,
FOLDED_NORMAL: FoldedNormalFitter,
GAMMA: GammaFitter,
GAUSSIAN: GaussianFitter,
NORMAL: GaussianFitter,
HALF_NORMAL: HalfNormalFitter,
LAPLACE: LaplaceFitter,
LOG_NORMAL: LogNormalFitter,
SKEW_NORMAL: SkewNormalFitter,
LINE: LineFitter,
}
[docs]
class MixedDataFitter:
r"""
Class to fit a mixture of different models to data.
:param x_values: The x-values for the data.
:param y_values: The y-values for the data.
:param model_list: List of models to fit (e.g., `LINE`, `GAUSSIAN`, `LOG_NORMAL`)
:param max_iterations: The maximum number of iterations for fitting procedure.
"""
def __init__(
self,
x_values: OneDArray,
y_values: OneDArray,
model_list: List[str],
fitter_dictionary: Optional[dict] = None,
model_dictionary: Optional[dict] = None,
max_iterations: int = 1000,
):
# Check if the deprecated parameter was used
if fitter_dictionary is not None:
warnings.warn(
message="`fitter_dictionary` is deprecated and will be removed in a future release. "
"Use `model_dictionary` instead.",
category=DeprecationWarning,
stacklevel=2,
)
x_values, y_values = sanity_check(x_values=x_values, y_values=y_values)
self.x_values: np.ndarray = x_values
self.y_values: np.ndarray = y_values
self.model_list = model_list
self.max_iterations = max_iterations
self.params: Any = None
self.covariance: Any = None
self.fitter_dict = fitter_dictionary or fitter_dict
self.fitter_dict = model_dictionary or fitter_dict
# self._validate_models()
self.model_function = self._create_model_function()
def __repr__(self):
return (
f"{self.__class__.__name__}(x_values={self.x_values}, y_values={self.y_values}, "
f"model_list={self.model_list}, max_iterations={self.max_iterations})"
)
[docs]
def _create_model_function(self) -> Callable:
"""
Creates a composite model function based on the specified models.
:return: A composite model for fitting.
"""
def _composite_model(x: np.ndarray, *params) -> np.ndarray:
"""
Compute the composite model.
Parameters
----------
x : np.ndarray
The x-values where the model is evaluated.
params : tuple
Parameters for the model components.
Returns
-------
y : np.ndarray
The computed y-values from the composite model.
"""
y = np.zeros_like(x, dtype=float)
param_index = 0
for model in self.model_list:
model_class = self._instantiate_class(model=model)
n_par = self._instantiate_n_par(model=model)
y += model_class.fitter(x=x, params=list(params[param_index : param_index + n_par]))
param_index += n_par
return y
return _composite_model
[docs]
def _expected_param_count(self) -> int:
"""
Calculates the expected number of parameters based on the model list.
:return: The number of parameters.
"""
count = 0
for model in self.model_list:
count += self._instantiate_n_par(model=model)
return count
[docs]
def _get_bounds(self):
"""
Sets the bounds for each parameter based on the model list.
:returns: Lower and upper bounds for the parameters.
"""
lower_bounds = []
upper_bounds = []
for model in self.model_list:
lb, ub = self._instantiate_bounds(model=model)
lower_bounds.extend(lb)
upper_bounds.extend(ub)
return np.array(lower_bounds), np.array(upper_bounds)
[docs]
def _instantiate_class(self, model: str):
try:
fitter_instance = self.fitter_dict[model](x_values=np.array([]), y_values=np.array([]))
except KeyError:
raise ValueError(f"Model '{model}' not recognized. Ensure it is defined in the fitter dictionary.")
return fitter_instance
[docs]
def _instantiate_n_par(self, model: str) -> int:
return self._instantiate_class(model).n_par
[docs]
def _instantiate_bounds(self, model: str) -> tuple[Sequence[float], Sequence[float]]:
return self._instantiate_class(model).fit_boundaries()
[docs]
def _params(self) -> np.ndarray:
r"""
Store the fitted parameters of the fitted model.
Returns
-------
np.ndarray
The parameters obtained after performing the fit.
Raises
------
RuntimeError
If the fit has not been performed yet (i.e., ``self.params`` is ``None``).
Notes
-----
This method assumes that the fitting process assigns values to ``self.params``.
"""
if self.params is None:
raise RuntimeError("Fit not performed yet. Call fit() first.")
return self.params
[docs]
def _plot_individual_fitter(self, plotter):
"""
Plot the individual fitters function.
:param plotter: The plotting axis object
"""
x = self.x_values
colors = plt.rcParams["axes.prop_cycle"].by_key()["color"][1:]
param_index = 0
for i, model in enumerate(self.model_list):
color = colors[i % len(colors)]
class_model = self._instantiate_class(model=model)
n_par = self._instantiate_n_par(model=model)
pars = self.params[param_index : param_index + n_par]
y_component = class_model.fitter(x=x, params=pars)
plot_xy(
x_data=x,
y_data=y_component,
x_label="",
y_label="",
plot_title="",
data_label=f"{model.capitalize()} {i + 1}({', '.join(self._format_param(i) for i in pars)})",
plot_dictionary=LinePlot(line_style="--", color=color),
axis=plotter,
)
param_index += n_par
[docs]
def _standard_errors(self) -> np.ndarray:
r"""
Store the standard errors of the fitted parameters.
Returns
-------
np.ndarray
An array containing the standard errors of the fitted parameters.
Raises
------
RuntimeError
If the fit has not been performed yet (i.e., ``self.covariance`` is ``None``).
"""
if self.covariance is None:
raise RuntimeError("Fit not performed yet. Call fit() first.")
return np.sqrt(np.diag(self.covariance))
[docs]
def fit(self, p0: Params_, frozen: Optional[Union[int, List[int]]] = None):
"""
Fit the data.
:param p0: Initial guess for the fitted parameters.
:type p0: Union[List[Tuple[int | float, ...]], np.ndarray]
:param frozen: Parameter number of list of parameter numbers to freeze the value of.
:type frozen: Union[int, List[int]]
:raises ValueError: If the length of the initial guess is not equal to the expected parameter count.
"""
p0_chain = p0.tolist() if isinstance(p0, np.ndarray) else p0
# flatten cannot always work here because the mixed fitter might contain a variable number of parameters
p0_chain = list(itertools.chain.from_iterable(p0_chain))
if len(p0_chain) != self._expected_param_count():
raise ValueError(
f"Initial parameters length {len(p0_chain)} does not match expected count "
f"{self._expected_param_count()}."
)
lb, ub = self._get_bounds()
if frozen:
if isinstance(frozen, int):
frozen = [frozen]
for par_num in frozen:
lb[par_num - 1] = p0_chain[par_num - 1] - epsilon
ub[par_num - 1] = p0_chain[par_num - 1] + epsilon
self.params, self.covariance, *_ = curve_fit(
f=self.model_function,
xdata=self.x_values,
ydata=self.y_values,
p0=np.array(p0_chain),
maxfev=self.max_iterations,
bounds=Bounds(lb=lb, ub=ub),
)
[docs]
def get_fitted_curve(self) -> np.ndarray:
"""
Gets the y-values from the fitted model.
:return: The y-values from the fitted model
:raises ValueError: If the model has not been fitted yet.
"""
if self.params is None:
raise RuntimeError("Fit not performed yet. Call fit() first.")
return self.model_function(self.x_values, *self.params)
[docs]
def get_model_parameters(self, model: Optional[str] = None, errors: bool = False):
"""
Extracts parameters (and error) values for a specific model, or for all models if no model is specified.
:param model: Model name to extract parameters for. If unspecified, extracts parameters for all models.
Defaults to ``None``.
:param errors: If ``True``, includes the errors in the returned output. Defaults to ``False``.
:return: A dictionary containing:
- "parameters": Nested dictionary of parameter values for each model if `get_errors` is True.
- "errors": Nested dictionary of errors for each model (if `get_errors=True`).
Otherwise, returns just the parameters directly.
"""
parameters = self._parameter_extractor(self.params)
errs = self._parameter_extractor(np.sqrt(np.diag(self.covariance)))
if not errors:
return parameters if model is None else parameters.get(model, [])
if model is None:
# Return a combined dictionary for all models
return {"parameters": parameters, "errors": errs}
# Prepare output for a specific model
output: dict = {"parameters": {}, "errors": {}}
keys = ["parameters", "errors"]
n_pars = self._instantiate_n_par(model=model)
for temp_, key in zip([parameters, errs], keys):
par_dict = temp_.get(model, [])
if n_pars == 2:
output[key] = par_dict
else:
output[key] = np.array_split(np.asarray(par_dict, dtype=float).flatten(), n_pars)
return output
[docs]
def get_value_error_pair(self, mean_values: bool = True, std_values: bool = False) -> np.ndarray:
r"""
Retrieve the value/error pairs for the fitted parameters.
This method provides the fitted parameter values and their corresponding standard errors as a combined array or
individually based on the input flags.
Parameters
----------
mean_values : bool, optional
If ``True``, return only the values of the fitted parameters.
Defaults to ``True``.
std_values : bool, optional
If ``True``, return only the standard errors of the fitted parameters.
Defaults to ``False``.
Returns
-------
np.ndarray
- If ``mean_values`` and ``std_values`` are both ``True``: A 2D array of shape (n_parameters, 2),
where each row is ``[value, error]``.
- If ``mean_values`` is ``True`` and ``std_values`` is ``False``: A 1D array of parameter values.
- If ``std_values`` is ``True`` and ``mean_values`` is ``False``: A 1D array of standard errors.
- If both flags are ``False``: An error message.
Raises
------
ValueError
If both ``mean_values`` and ``std_values`` are ``False``.
"""
pairs = np.column_stack([self._params(), self._standard_errors()])
if mean_values and std_values:
return pairs
elif mean_values:
return pairs[:, 0]
elif std_values:
return pairs[:, 1]
else:
raise ValueError("Either 'mean_values' or 'std_values' must be True.")
[docs]
def plot_fit(
self,
show_individuals: bool = False,
x_label: Optional[str] = None,
y_label: Optional[str] = None,
data_label: Optional[str] = None,
fit_label: Optional[str] = None,
title: Optional[str] = None,
axis: Optional[Axes] = None,
):
"""
Plot the fitted models.
Parameters
----------
show_individuals: bool, optional
Whether to show individually fitted models or not.
x_label: str, optional
The label for the x-axis.
y_label: str, optional
The label for the y-axis.
title: str, optional
The title for the plot.
data_label: str, optional
The label for the data.
fit_label: str, optional
The label for the fitted model.
axis: Axes, optional
Axes to plot instead of the entire figure. Defaults to None.
Returns
-------
plotter
The plotter handle for the drawn plot.
"""
return _plot_fit(
x_values=self.x_values,
y_values=self.y_values,
parameters=self.params,
n_fits=len(self.model_list),
class_name=self.__class__.__name__,
_n_fitter=self.model_function,
_n_plotter=self._plot_individual_fitter,
show_individuals=show_individuals,
x_label=x_label,
y_label=y_label,
title=title,
data_label=data_label,
fit_label=fit_label,
axis=axis,
)