A Comparative Machine Learning Analysis and Hyperparameter Optimization Applied on Chronic Kidney Disease Prediction
Yazarlar (2)
Mariam Kili Bechır
Dr. Öğr. Üyesi Ferhat ATASOY Karabük Üniversitesi, Türkiye
| Bildiri Türü |
Tebliğ/Bildiri
|
Bildiri Dili | İngilizce |
| Bildiri Alt Türü | Özet Metin Olarak Yayınlanan Tebliğ (Ulusal Kongre/Sempozyum) | ||
| Bildiri Niteliği | Alanında Hakemli Ulusal Kongre/Sempozyum | ||
| Kongre Adı | 30. Ulusal Ergonomi Kongresi | ||
| Kongre Tarihi | 10-10-2024 / | ||
| Basıldığı Ülke | Türkiye | Basıldığı Şehir | Karabük |
| Bildiri Linki | https://uek30.karabuk.edu.tr/yuklenen/dosyalar/12613122024164527.pdf | ||
| UAK Araştırma Alanları |
Gömülü Sistemler
|
||
| Özet |
| Global health burden of Chronic Kidney Disease (CKD) necessitates early detection and accurate prediction. Machine Learning (ML) offers promising tools for CKD prediction using patient data. This study conducts a comparative analysis of various ML algorithms including ensemble techniques for CKD prediction. The analysis involves data cleaning, feature engineering, and exploration to understand the relationships between features and the data distribution. We employed various ML models, including K-Nearest Neighbors (KNN), Logistic Regression, Support Vector Machines (SVM), Decision Trees, Random Forests, XGBoost, and Gradient Boosting. We evaluated these models using metrics like Precision, Recall, F1-score, accuracy, ROC AUC score and confusion matrix to assess their performance in classifying patients with and without CKD. The findings demonstrate that ensemble methods, particularly Random Forest, Gradient Boosting and XGBoost, achieve the highest accuracy after hyperparameter optimization. Logistic Regression also performs well, suggesting a potentially linear relationship between features and the target variable. Conversely, KNN and SVM show lower accuracy, indicating their limitations in this specific classification task. Our comparative analysis aligns with existing research, highlighting the effectiveness of ML, particularly ensemble methods with hyperparameter tuning, for CKD prediction. |
| Anahtar Kelimeler |
