"Vocabulary: Classical Machine Learning"

Finding structure in unlabeled data Training on historical data with known outcomes Exploring data for anomaly detection Grouping similar records

The fraction of items flagged that were actually correct The ratio of true positives to all actual positives The measure of the model's ability to recall all important items The percentage of items that were missed by the model

It generalizes well to new data It performs similarly on training and test data It has memorized the training data, failing to perform on new data It has a high precision and high recall rate

An input variable for the model A tool for data structuring The known outcome associated with historical records The process of grouping records by similarity