<![CDATA[Diabetes mellitus is a chronic disease influenced by multiple interrelated factors, making the identification of causal relationships among risk factors important for understanding the mechanisms underlying diabetes. This study aims to analyze the causal structure of diabetes risk factors using the Greedy Equivalence Search (GES) algorithm. The dataset consists of nine variables, namely gender, age, hypertension, heart disease, smoking history, BMI, HbA1c level, blood glucose level, and diabetes. The research stages include literature review, data preprocessing, Formation of True DAG, causal modeling using GES, and evaluation by comparing the estimated structure with the True Directed Acyclic Graph (True DAG). The implementation was conducted using the R programming language with the pcalg package and the GaussL0penObsScore scoring function. The results show that GES successfully identified 14 out of 15 edges in the True DAG, with 8 false positives and 1 false negative. The evaluation yielded a precision of 63.64%, recall of 93.33%, F1-score of 75.4%, and accuracy of 87.50%. The GES-derived structure also identified several relationships relevant to diabetes, particularly involving HbA1c level, blood glucose level, BMI, age, and smoking history. These findings indicate that GES can be used to identify the causal structure of diabetes risk factors based on simulational data.]]>
Copyrights © 2026
