Ordinal Classification / Ordinal Regression class

OrdinalClassficiation.py

from sklearn.base import BaseEstimator
import numpy as np
class OrdClass(BaseEstimator):
  """
  Helper class that solves ordinal classification (classes that have an order to them eg cold,warm,hot)
  
  Paper (2001): https://www.cs.waikato.ac.nz/~eibe/pubs/ordinal_tech_report.pdf
  Fixed source from: https://github.com/garyongguanjie/Ordinal-Classifier
  """
  def __init__(self,classifier=None,clf_args=None):
    """
    y needs to be a number that start from 0 and increments by 1
    classifier object needs to be able to return a probability
    """
    self.classifier = classifier
    self.clfs = []
    self.clf_args = clf_args
    self.final_prob = None
  
  def fit(self,X,y,**fit):
    self.X = X
    self.y = y
    import copy
    no_of_classifiers = int(np.max(self.y)) #since y starts from 0
    self.classes_ = list(range(no_of_classifiers+1))
    for i in range(no_of_classifiers):
        # make a copy of y because we want to change the values of y
        copy_y = np.copy(self.y)
        # make a binary classification here
        copy_y[copy_y<=i] = 0
        copy_y[copy_y>i] = 1
        classifier = copy.deepcopy(self.classifier)
        classifier.fit(self.X,copy_y,**fit)
        self.clfs.append(classifier)
    return self

  def predict_proba(self,test):
    prob_list = []
    final_prob = []
    length = len(self.clfs)
    for clf in self.clfs:
          prob_list.append(clf.predict_proba(test)[:,1])
    for i in range(length+1):
        if i == 0:
            final_prob.append(1-prob_list[i])
        elif i == length:
            final_prob.append(prob_list[i-1])
        else:
            final_prob.append(prob_list[i-1]-prob_list[i])
    answer = np.array(final_prob).transpose()
    self.final_prob= answer
    return answer

  def predict(self,test):
    self.predict_proba(test)
    return np.argmax(self.final_prob,axis=1)

  def score(self,X,y,sample_weight=None):
    return accuracy_score(y, self.predict(X), sample_weight=sample_weight)