Infer.NET user guide : Learners : Bayes Point Machine classifiers : The Learner API : Mappings

Evaluation Data Format Mapping

An easy way to assess the performance of a classifier is to use an evaluator. The evaluator reads the ground truth labels for some instances of interest (validation or test set) via a mapping which implements the IClassifierEvaluatorMapping interface. Since an evaluator should be independent of the concrete data formats required by specific classifier implementations such as the Bayes Point Machine, IClassifierEvaluatorMapping essentially declares the generic standard data format mapping of the IClassifierMapping interface, just without the GetFeatures method. Predictions are input arguments to the evaluation methods and do not get accessed via the mapping.

A concrete implementation of the IClassifierEvaluatorMapping interface can be defined based on a given standard data format mapping. In fact, there is an extension method, ForEvaluation, to do just that. It takes the given standard data format mapping and returns the corresponding classifier evaluator mapping, essentially producing what is called a chained mapping.