Aan de slagGa gratis aan de slag

Overfitting and underfitting

Interpreting model complexity is a great way to evaluate supervised learning performance. Your aim is to produce a model that can interpret the relationship between features and the target variable, as well as generalize well when exposed to new observations.

The training and test sets have been created from the churn_df dataset and preloaded as X_train, X_test, y_train, and y_test.

In addition, KNeighborsClassifier has been imported for you along with numpy as np.

Deze oefening maakt deel uit van de cursus

Supervised Learning with scikit-learn

Cursus bekijken

Oefeninstructies

  • Create neighbors as a numpy array of values from 1 up to and including 12.
  • Instantiate a KNeighborsClassifier, with the number of neighbors equal to the neighbor iterator.
  • Fit the model to the training data.
  • Calculate accuracy scores for the training set and test set separately using the .score() method, and assign the results to the train_accuracies and test_accuracies dictionaries, respectively, utilizing the neighbor iterator as the index.

Praktische interactieve oefening

Probeer deze oefening eens door deze voorbeeldcode in te vullen.

# Create neighbors
neighbors = np.arange(____, ____)
train_accuracies = {}
test_accuracies = {}

for neighbor in neighbors:
  
	# Set up a KNN Classifier
	knn = ____(____=____)
  
	# Fit the model
	knn.____(____, ____)
  
	# Compute accuracy
	train_accuracies[____] = knn.____(____, ____)
	test_accuracies[____] = knn.____(____, ____)
print(neighbors, '\n', train_accuracies, '\n', test_accuracies)
Code bewerken en uitvoeren