CLASSIFICATIONS METHODS AND APPLICATIONS IN HEALTH RESEARCH
Project no. 32299 · Status: Kapanmış (Yayın bekleniyor) · 2024
Summary
<p>In this study, data mining classification methods were discussed, specifically focusing on the algorithms J48, k-nearest neighbors (KNN), logistic regression (LR), Naive Bayes (NB), random forest (RF), and support vector machines (SVM). Additionally, information was provided on the software tools used in the study, namely WEKA, RapidMiner, and Orange. The aim of the study is to compare the performance of different classification algorithms using various software tools. The application part of the study was conducted using the UCTH Breast Cancer dataset obtained from the Mendeley Data repository. The dataset consists of observations from 213 breast cancer patients, 120 diagnosed with benign tumors and 93 with malignant tumors. According to the results obtained, although the software tools produced relatively similar outcomes, the highest accuracy rate was achieved using the KNN algorithm implemented in WEKA, with a value of 95.71%. The same algorithm in WEKA also yielded the highest values in terms of sensitivity (97.6%), F-measure (96.4%), and kappa statistic (91.1%). In terms of precision (97.3%) and specificity (96.4%), the Random Forest algorithm in WEKA showed the best performance. Regarding the area under the ROC curve (AUC), the highest result was obtained with Logistic Regression in WEKA, with a value of 98.1%.</p><p><strong>Keywords:</strong> Data Mining, Random Forest, Naive Bayes, Logistic Regression, J48, KNN, Support Vector Machines.</p>