In Reasoned Regression, you play as a data scientist who uses analytic methods to generate properties about the data under study, and Prolog to reason about the data via the generated properties. Solve puzzles by partitioning data at random along explanatory variables and fitting linear or isotonic regressions to each partition segment.