from typing import List
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
from ..impute import get_observed_mod_indicator
from ..utils import check_Xs
[docs]
class SimpleModImputer(BaseEstimator, TransformerMixin):
r"""
Fill incomplete samples of a dataset using a specified method.
Parameters
----------
value : str, optional (default='mean')
The method to use for filling missing modalities. Possible values:
- 'mean': replace missing samples with the mean of each feature in the corresponding modality.
- 'zeros': replace missing samples with zeros.
Attributes
----------
features_mod_mean_list_ : array-like of shape (n_mods,)
The mean value of each feature in the corresponding modality, if value='mean'
Example
--------
>>> import numpy as np
>>> import pandas as pd
>>> from imml.impute import SimpleModImputer
>>> from imml.ampute import Amputer
>>> Xs = [pd.DataFrame(np.random.default_rng(42).random((20, 10))) for i in range(3)]
>>> amp = Amputer(p=0.3, random_state=42)
>>> Xs = amp.fit_transform(Xs)
>>> transformer = SimpleModImputer(value = 'mean')
>>> transformer.fit_transform(Xs)
"""
def __init__(self, value : str = 'mean'):
values = ['mean', 'zeros']
if value not in values:
raise ValueError(f"Invalid value. Expected one of: {values}")
self.value = value
[docs]
def fit(self, Xs : List, y=None):
r"""
Fit the transformer to the input data.
Parameters
----------
Xs : list of array-likes objects
- Xs length: n_mods
- Xs[i] shape: (n_samples, n_features_i)
A list of different modalities.
y : Ignored
Not used, present here for API consistency by convention.
Returns
-------
self : returns an instance of self.
"""
Xs = check_Xs(Xs, ensure_all_finite='allow-nan')
if self.value == "mean":
self.features_mod_mean_list_ = [np.nanmean(X, axis=0) for X in Xs]
elif self.value == "zeros":
pass
return self
def simple_mod_imputer(Xs : list, y = None, value : str = 'mean'):
r"""
Transform the input data by filling missing samples.
Parameters
----------
Xs : list of array-likes objects
- Xs length: n_mods
- Xs[i] shape: (n_samples, n_features)
A list of different modalities.
y : Ignored
Not used, present here for API consistency by convention.
value : str, optional (default='mean')
The method to use for filling missing modalities. Possible values:
- 'mean': replace missing samples with the mean of each feature in the corresponding modality.
- 'zeros': replace missing samples with zeros.
Returns
-------
transformed_Xs : list of array-likes objects, shape (n_samples, n_features_i)
The transformed data with filled missing samples.
"""
Xs = check_Xs(Xs, ensure_all_finite='allow-nan')
pandas_format = isinstance(Xs[0], pd.DataFrame)
if pandas_format:
samples = Xs[0].index
features = [X.columns for X in Xs]
dtypes = [X.dtypes.to_dict() for X in Xs]
Xs = [X.values for X in Xs]
observed_mod_indicator = get_observed_mod_indicator(Xs=Xs)
n_samples = len(observed_mod_indicator)
transformed_Xs = []
for X_idx, X in enumerate(Xs):
n_features = X.shape[1]
if value == "mean":
features_mod_mean = np.nanmean(X, axis=0)
transformed_X = np.tile(features_mod_mean, (n_samples, 1))
elif value == "zeros":
transformed_X = np.zeros((n_samples, n_features))
else:
raise ValueError(f"Invalid value. Expected one of: ['mean', 'zeros']")
transformed_X[observed_mod_indicator[:, X_idx]] = X[observed_mod_indicator[:, X_idx]]
if pandas_format:
transformed_X = pd.DataFrame(transformed_X, index=samples, columns=features[X_idx])
transformed_X = transformed_X.astype(dtypes[X_idx])
transformed_Xs.append(transformed_X)
return transformed_Xs