File heuristic_precision.hpp

namespace seco
class PrecisionConfig : public seco::IHeuristicConfig
#include <heuristic_precision.hpp>

Allows to configure a heuristic that measures the fraction of correctly predicted labels among all labels that are covered by a rule.

This heuristic is equivalent to RIPPER’s pruning heuristic (“Fast Effective Rule Induction”, Cohen 1995, see https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=6c34b7b0441bff66cce2418d36acfd9776ad7bd2). A proof is provided in the paper “Roc ‘n’ Rule Learning — Towards a Better Understanding of Covering Algorithms”, Fürnkranz, Flach 2005 (see https://link.springer.com/content/pdf/10.1007/s10994-005-5011-x.pdf).

Public Functions

virtual std::unique_ptr<IHeuristicFactory> createHeuristicFactory() const override

Creates and returns a new object of type IHeuristicFactory according to the specified configuration.

Returns:

An unique pointer to an object of type IHeuristicFactory that has been created