"""
Homogeneous Linear Model similar to Oh et al. 2014.
"""
# Authors: Joseph Knox <josephk@alleninstitute.org>
# License: Allen Institute Software License
from sklearn.utils.validation import check_is_fitted
from ...regressors import NonnegativeLinear
from .subset_selection import backward_subset_selection_conditioning
[docs]class HomogeneousModel(NonnegativeLinear):
"""Homogeneous model similar to Oh et al. 2014.
Implements the Homogeneous model for fitting nonnegative weights at the
regional level. There is an additional constraint on the features to
ensure the regression is well conditioned.
Parameters
----------
kappa : float, optional, default: 1000
The maximum condition number allowed for input data arrays.
Attributes
----------
columns_ : array
The features that are included in the regression after the conditioning
has occurred.
coef_ : array, shape (n_targets, n_features)
The model weights.
Examples
--------
>>> from mcmodels.core import VoxelModelCache, RegionalData
>>> from mcmodels.models import HomogeneousModel
>>> # get data with whcich to fit model
>>> cache = VoxelModelCache()
>>> voxel_data = cache.get_experiment_data()
>>> regional_data = RegionalData.from_voxel_data(voxel_data)
>>> # fit model
>>> reg = HomogeneousModel()
>>> reg.fit(regional_data.injections, regional_data.projections)
HomogeneousModel(kappa=1000)
References
----------
Oh et al. 2014. A mesoscale connectome of the mouse brain. Nature,
508(7495), 207-214. doi: 10.1038/nature13186
"""
[docs] def __init__(self, kappa=1000):
self.kappa = kappa
[docs] def fit(self, X, y, sample_weight=None):
"""Fit HomogeneousModel.
Parameters
----------
X : array, shape (n_samples, n_features)
Training data.
y : array, shape (n_samples, n_features)
Target values.
Returns
-------
self : returns an instance of self
"""
X, columns = backward_subset_selection_conditioning(X, self.kappa)
self.columns_ = columns
return super(HomogeneousModel, self).fit(X, y, sample_weight=sample_weight)
[docs] def predict(self, X):
"""Predict using the HomogeneousModel.
Parameters
----------
X : array, shape (n_samples, n_features)
Training data.
Returns
-------
C : array, shape (n_samples,) or (n_samples, n_targets)
Returns predicted values.
"""
check_is_fitted(self, "columns_")
y_pred = super(HomogeneousModel, self).predict(X[:, self.columns_])
return y_pred
@property
def weights(self):
"""Convenience property for pulling out regional matrix."""
check_is_fitted(self, ["coef_"])
return self.coef_.T