Which of these is NOT a common application of unsupervised machine learning?

Unsupervised machine learning focuses on identifying patterns and structures in data without the use of labeled outputs. This type of learning is commonly applied in situations where the goal is to explore the underlying structure of data and to group observations based on their similarities.

Market segmentation involves clustering customer data into distinct groups based on similar characteristics, making it a prime example of unsupervised learning. Similarly, image compression often utilizes unsupervised techniques to reduce the amount of data required to represent an image, enabling efficient storage and transmission.

On the other hand, spam detection is primarily a supervised learning task. It requires a labeled dataset where emails are pre-classified as either "spam" or "not spam." The model learns from these labeled examples to predict the category of new, unseen emails.

Customer churn prediction also relies on supervised learning, as it involves using historical data about customers who have left (churned) along with other labeled information to predict future churn.

Therefore, spam detection is not a common application of unsupervised machine learning, as it relies on prior knowledge of labels to train the predictive model.