Classification Model Selection Using a Mahalanobis Based Multi Criteria Decision Making Metric Mahalanobis Tabanli Çok Kriterli Karar Verme Metri?gi ?Ile Siniflandirma Modeli Seçimi
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11637074
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: classifier selection, machine learning, Mahalanobis distance, multi criteria decision making, wisconsin breast cancer
- Yıldız Teknik Üniversitesi Adresli: Hayır
Özet
Selecting the most suitable classifier in machine learning requires evaluating multiple performance metrics such as accuracy, precision, sensitivity, and F1 score together. However, examining these metrics separately makes comparison difficult when different models excel in different criteria. This study proposes a Mahalanobis-based composite metric that considers both the correlation between performance metrics and model stability across cross-validation folds within a multi criteria decision making approach. The proposed method represents each model as a performance vector, calculates the weighted distance from the ideal values of the metrics, and integrates stability by adding the standard deviation as a penalty score. The results of the study demonstrate that the approach is simple, understandable, and applicable for multi-criteria model evaluation.