Document worth reading: “Performance Metrics (Error Measures) in Machine Learning Regression, Forecasting and Prognostics: Properties and Typology”

Performance metrics (error measures) are essential parts of the evaluation frameworks in diversified fields. The intention of this analysis was to overview of a variety of effectivity metrics and approaches to their classification. The most essential goal of the analysis was to develop a typology that will help to boost our knowledge and understanding of metrics and facilitate their alternative in machine finding out regression, forecasting and prognostics. Based on the analysis of the development of fairly a couple of effectivity metrics, we recommend a framework of metrics which contains 4 (4) lessons: main metrics, extended metrics, composite metrics, and hybrid items of metrics. The paper acknowledged three (3) key parts (dimensions) that determine the development and properties of main metrics: strategy of determining degree distance, strategy of normalization, strategy of aggregation of degree distances over a data set. Performance Metrics (Error Measures) in Machine Learning Regression, Forecasting and Prognostics: Properties and Typology