1. Supervised learning is used to predict outcomes from inputs using input-output examples provided by a supervisor to train a model. 2. Unsupervised learning is used to extract knowledge from input data without supervision by looking for patterns in the data. 3. A decision tree model can be built using a student performance dataset to predict academic performance of new students based on attributes like study time and absences. The machine learning software builds the decision tree by analyzing the training data.