Joint classifier for FeatureUnion in the pipeline

joint_classifier.py

'''
Created on June 26, 2013

@author: dashesy

Purpose: glue the classifier wire logic to have multiple 
         classifiers work jointly with different groups of X,y
         Each group could accept different number os samples or features

'''
import numpy as np
import warnings
from scipy import linalg

from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.utils import array2d

class JointClassifier(BaseEstimator, TransformerMixin):
    '''
    Glue logic for FeatureUnion to accept different X,y 
    If joint group numbers match acts as a connected wire, otherwise as a buffered wire
    
    Parameters
    ----------
    group : int, 
        Transformation group number (0-based).
        Default is 1 (i.e. the second group)
        
    copy : bool
        If False, data passed to fit are overwritten
        self._n_samples = n_samples
        self._n_features = n_features        
    Attributes
    ----------
    `_X` : array, shape=[n_samples, n_features]
        Previous Training data.
    `_y` : array-like, shape = [n_samples]
        Previous Target vector relative to X
    '_X_transformed' : array, shape=[n_samples, n_features]
        Previous transformation result
    '''
    
    def __init__(self, joint_group = 1, copy=True):
        self.copy = copy
        self.joint_group = joint_group

    def fit(self, X, y, **params):
        """Fit the model according to the given training data

        Parameters
        ----------
        X: array-like, shape (n_samples, n_features)
            Training data, where n_samples in the number of samples
            and n_features is the number of features.
            
        y : array-like, shape = [n_samples]
            Target vector relative to X
        
        joint_group: Transformation group number (0-based).
            Groups must match in order to fit or transform
            Default is 0 (i.e. the first group)

        Returns
        -------
        self : object
            Returns the instance itself.
        """
        
        joint_group = 0
        if params.has_key('joint_group'):
            joint_group = params['joint_group']
        
        if joint_group == self.joint_group: 
            self._fit(X, y)
        return self

    def fit_transform(self, X, y, **params):
        
        """Fit the model with X and apply the dimensionality reduction on X.

        Parameters
        ----------
        X : array-like, shape (n_samples, n_features)
            Training data, where n_samples in the number of samples
            and n_features is the number of features.

        y : array-like, shape = [n_samples]
            Target vector relative to X
            
        joint_group: Transformation group number (0-based).
            Groups must match in order to fit or transform
            Default is 0 (i.e. the first group)

        Returns
        -------
        X_new : array-like, shape (n_samples, n_axis)

        """
        
        joint_group = 0
        if params.has_key('joint_group'):
            joint_group = params['joint_group']
        
        if joint_group == self.joint_group: 
            self._fit(X, y)
            
        return self.transform(X, **params)
            
        
    
    def _fit(self, X, y = None):
        '''
        Fit the model with X
        '''

        X = array2d(X)
        
        # Buffer X,y
        self._X = X
        self._y = y

    
    def transform(self, X, **params):
        """Apply the dimensionality reduction on X.

        Parameters
        ----------
        X : array-like, shape (n_samples, n_features)
            New data, where n_samples in the number of samples
            and n_features is the number of features.
            
        joint_group: Transformation group number (0-based).
            If joint groups matche this transformation acts as a connected wire, otherwise as a buffered wire for previous transformation
            Default is 0 (i.e. the first group)

        Returns
        -------
        X_new : array-like, shape (n_samples, n_axis)

        """
        
        joint_group = 0
        if params.has_key('joint_group'):
            joint_group = params['joint_group']

        X = array2d(X)
        n_samples, n_features = X.shape
        
        if joint_group != self.joint_group:
            # Model must be transformed once before with matching joint group
            return self._X_transformed
        
        self._X_transformed = X
        return X