Create a Simple Estimator

In this chapter we will cover the following recipes:

• Creating a simple estimator

Introduction

We are going to make a custom estimator with scikit-learn. We will take traditional statistical math and programming and turn it into machine learning. You are able to turn any statistics into machine learning by using scikit-learn's powerful cross-validation capabilities.

Create a simple estimator

We are going to do some work towards building our own scikit-learn estimator. The custom scikit-learn estimator consists of at least three methods:

• An __init__ initialization method: This method takes as input the estimator's parameters
• A fit method: This trains the estimator
• A predict method: This method performs a prediction on unseen data

Schematically, the class looks like this:

#Inherit from the classes BaseEstimator, ClassifierMixin
class RidgeClassifier(BaseEstimator, ClassifierMixin):
 
     def __init__(self,param1,param2):
          self.param1 = param1
          self.param2 = param2

     def fit(self, X, y = None):
          #do as much work as possible in this method
          return self
 
     def predict(self, X_test):
          #do some work here and return the predictions, y_pred
          return y_pred 

...