"""Defines the GaussianNB model class."""
from typing import Any, Dict, List, Optional
import numpy as np
import pandas as pd
from ml_grid.util import param_space
from ml_grid.util.global_params import global_parameters
from sklearn.naive_bayes import GaussianNB
from skopt.space import Categorical
print("Imported gaussiannb class")
[docs]
class GaussianNBWrapper(GaussianNB):
"""A wrapper for GaussianNB to handle integer-mapped priors for Bayesian search."""
[docs]
def set_params(self, **params: Any) -> "GaussianNBWrapper":
"""Sets the parameters of the estimator.
This method intercepts the 'priors' parameter if it's an integer index
and maps it to the corresponding list of prior probabilities before
passing it to the parent's set_params method.
Args:
**params (Any): Estimator parameters.
Returns:
GaussianNBWrapper: The instance with updated parameters.
"""
prior_mapping: Dict[int, Optional[List[float]]] = {
0: None, # Default priors (based on the class distribution in the dataset)
1: [0.5, 0.5], # Equal probabilities
2: [0.6, 0.4], # Slight imbalance favoring class 0
3: [0.4, 0.6], # Slight imbalance favoring class 1
4: [0.7, 0.3], # Moderate imbalance favoring class 0
5: [0.3, 0.7], # Moderate imbalance favoring class 1
6: [0.8, 0.2], # Strong imbalance favoring class 0
7: [0.2, 0.8], # Strong imbalance favoring class 1
8: [0.9, 0.1], # Extreme imbalance favoring class 0
9: [0.1, 0.9], # Extreme imbalance favoring class 1
}
if "priors" in params:
priors_idx = params.pop("priors")
params["priors"] = prior_mapping[priors_idx]
return super().set_params(**params)
[docs]
class GaussianNB_class:
"""GaussianNB classifier with hyperparameter tuning support."""
def __init__(
self,
X: Optional[pd.DataFrame] = None,
y: Optional[pd.Series] = None,
parameter_space_size: Optional[str] = None,
):
"""Initializes the GaussianNB_class.
Args:
X (Optional[pd.DataFrame]): Feature matrix for training.
Defaults to None.
y (Optional[pd.Series]): Target vector for training.
Defaults to None.
parameter_space_size (Optional[str]): Size of the parameter space for
optimization. Defaults to None.
"""
global_params = global_parameters
if global_params.bayessearch is False:
self.algorithm_implementation = GaussianNB()
else:
self.algorithm_implementation = (
GaussianNBWrapper()
) # Wrapper necessary for passing priors to bayescv
[docs]
self.method_name = "GaussianNB"
[docs]
self.parameter_vector_space = param_space.ParamSpace(parameter_space_size)
# print(self.parameter_vector_space)
if global_params.bayessearch:
# For BayesSearchCV, use distributions from skopt.space
self.parameter_space = {
'var_smoothing': self.parameter_vector_space.param_dict.get("log_small"),
'priors': Categorical([0, 1, 2]) # Integer mapping
}
# Log parameter space for verification
print(f"Parameter Space: {self.parameter_space}")
else:
# For traditional grid search, use lists
self.new_list = list(
self.parameter_vector_space.param_dict.get("log_small")
).copy()
self.new_list.append(1e-9) # Add the small value explicitly
self.parameter_space = {
"priors": [
None,
[0.1, 0.9],
[0.9, 0.1],
[0.7, 0.3],
[0.3, 0.7],
[0.5, 0.5],
[0.6, 0.4],
[0.4, 0.6],
], # Enumerates possible values as a list
"var_smoothing": self.new_list,
}
return None
# print("init log reg class ", self.parameter_space)