Building justified classification models for decision making and forecasting

Ponomareva K. A., Stupina A. A., Fedorova A. V., Korpacheva L. N.

Classification is an important task of data mining, where the value of a discrete (dependent) variable is predicted based on the values of some independent variables. Classification models must provide correct predictions for new data instances. This article focuses on the key requirements for such models in any field where the model must be tested before it can be implemented. The main requirements for classification models are clarity and validity, which reflects the model's compliance with existing knowledge of the subject area under consideration. Providing clarity and validity of classification models contributes to their practical application in areas where previously such models were considered too theoretical and incomprehensible. So, a classification model that is accurate, understandable, and efficient is defined as acceptable for implementation.

Keywords:data mining, model productivity, classification, comprehensibility, justifiability, forecasting, medical diagnostics.