A Study on the Selection of Performance Evaluation Metrics for Machine Learning Methods

As a cornerstone technology in the realm of artificial intelligence, machine learning exhibits substantial potential for application across diverse disciplines. The aim of this study is to investigate the method for evaluating the effectiveness of machine learning. The current machine learning performance evaluation indexes are first analyzed, including Precision, Recall, F1 Score, Precision-Recall <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Curve, Optimal Threshold, F</sup>β Score, Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Coefficient of Determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>). Secondly, a multi-dimensional combat damage evaluation model was established and its learning effect evaluation index system was proposed. The results show that the selected evaluation indexes of machine learning performance can effectively evaluate the combat damage of UAV in complex environments, and the applicability and accuracy of the model have been verified. This study provides a new analysis method for combat damage assessment in manned/unmanned coordinated combat, enhancing the operational efficiency and safety of UAVs holds great significance.