A community-theater box office wants last season’s no-shows sorted into “likely late” versus “likely cancel” after a person picks columns such as weekday and weather. Which AI approach does this describe?

Classical machine learning fits the community-theater box-office task when historical no-shows are prepared, features such as weekday and weather are chosen by a person, and a training pass produces categories like “likely late” versus “likely cancel.” The defining pattern is data preparation, hand-selected columns, and supervised learning—not an editable rule card and not content generation. Civic box-office teams should recognize that labeling season history and picking predictive columns is classical ML work.

Symbolic AI fails because it would apply a fixed if-then table a clerk edits by hand rather than train on prepared feature columns. Generative AI fails because writing new reminder scripts from a user story creates content; it does not classify historical no-shows after feature selection. Deep learning fails the scenario’s stated method: the person picks columns, whereas deep learning is designed to learn features automatically from large messy inputs such as raw seat photographs.

Exam caveat: classical ML is identified by prepared data, chosen features, and training—not by the mere presence of a theater or ticket system. Vendor pitches that skip feature selection while claiming classical ML often actually mean deep learning or GenAI.

Operational check: list the selected feature columns; confirm a labeled training set exists; verify the model outputs category scores rather than drafted scripts; record in the municipal test notes that human feature choice was the discriminator versus deep learning.