In supervised learning, the system is trained on a labelled dataset with a known answer. This serves as a “supervisor” for training the model, providing an answer key for the algorithm to assess its correctness on training data. This is used to forecast values for unknown or future data. Predicting product sales prices, how much loan to award, churn prediction, and Titanic survivors are all examples of supervised learning.
The model infers the hidden structure, pattern, or structure from unlabeled data in unsupervised learning. In such data, there is no response or goal variable to supervise the analysis from what is correct or incorrect. The system attempts to recognize the pattern and responds accordingly. In the absence of the intended result, clustering is used to categorize or segment the data. The algorithm learns to tell the difference between a human face and a horse or cat’s face. Customer segmentation, picture segmentation, market basket analysis, delivery store optimization, and detecting accident-prone locations are all examples of unsupervised learning.