Decision Tree
A flowchart classifier that splits the data with if/else tests on features until the leaves are mostly one class. At each node it picks the feature and threshold that minimise the weighted Gini Impurity of the children.
When to use
Its strengths are that it is interpretable and needs no scaling; its weaknesses are instability and overfitting, so limit it with max_depth or min_samples_leaf, or ensemble it as a Random Forest. On the Yalla Fitness churn data it scored 78% accuracy.
See DS Lec 3.