<![CDATA[The increase in online lending distribution in Indonesia in 2024 was not accompanied by a uniform level of credit risk across regions. This study aims to categorize online lending regions in Indonesia based on outstanding values and 90-day default rates (TWP90) using a quantitative approach based on the K-Means algorithm. Secondary data from all provinces was analyzed using RapidMiner and evaluated using the Davies–Bouldin Index (DBI). The test results showed a DBI of 0.746 at K=2, 0.376 at K=3, and 0.564 at K=4. Although K=2 yielded the lowest DBI, the K=3 model was chosen because it provided a more informative and policy-relevant risk classification. The clustering resulted in three risk clusters: Low Risk, with outstanding values and TWP90 below average; Medium Risk, with values above average; and High Risk, characterized by a very high TWP90 level despite relatively low outstanding values. These findings confirm the effectiveness of K-Means in mapping online lending risks based on regions and support more precise credit monitoring. Keywords: online loans, K-Means clustering, default risk, outstanding, TWP90.]]>
Copyrights © 2026
