Evaluation Metrics in Learning Systems: A Survey

Learning systems have played a significant role in classification, Learning sys tems have shaped classification, clustering, regression, and pattern learning problems within the past few decades. Machine learning and deep learning continue to churn out newer models that have transformed the areas of natural language processing, computer vision, speech recognition, recommender sys tems, among others. The evaluation metrics serve as verification systems and are a tool to benchmark the models presented. Depending on the field of use and kind of application, different metrics have been put forth in the literature. They form the common language between researchers. Commercial projects use these metrics to help companies evaluate the value of a model from both an economic and practical point of view. Moreover, in the absence of standards for measurement and evaluation criteria, it becomes almost impossible to compare the results obtained from similar models. This study aims to classify different metrics in learning systems. Therefore, approximately 250 evaluation metrics in text mining, clustering, image mining, and signal analysis have been presented.