Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/160218| Title: | Perbandingan Analisis K-median dan Fuzzy C-means Clustering pada Asuransi Kesehatan |
| Other Titles: | Comparison of K-median and Fuzzy C-means Clustering Analysis in Health Insurance |
| Authors: | Budiarti, Retno Ardana, Ngakan Komang Kutha Mertorini, Winda |
| Issue Date: | 2024 |
| Publisher: | IPB University |
| Abstract: | Penelitian ini bertujuan untuk membandingkan metode k-median dan fuzzy c-means (FCM) clustering dalam mengelompokkan data pelanggan asuransi kesehatan. Data yang digunakan meliputi usia, BMI, gaya hidup, dan biaya kesehatan pelanggan di Amerika Serikat. Uji multikolinearitas menunjukkan tidak adanya masalah multikolinearitas, sehingga semua variabel dapat digunakan dalam analisis clustering. Penentuan jumlah cluster optimal dilakukan menggunakan indeks silhouette coefficient, yang menunjukkan bahwa dua cluster adalah pilihan terbaik untuk kedua metode. Berdasarkan evaluasi Davies-Bouldin Index (DBI), metode k-median terbukti lebih unggul dibandingkan FCM karena menghasilkan cluster yang lebih kompak dan terpisah dengan baik. Hasil clustering k-median juga menunjukkan pemisahan yang lebih jelas berdasarkan variabel kategorikal seperti gaya hidup. Keunggulan ini membuktikan bahwa metode k-median memiliki karakteristik cluster yang lebih relevan dengan perbedaan signifikan pada usia, BMI, kebiasaan merokok, dan biaya kesehatan/This study aims to compare the k-median and fuzzy c-means (FCM) clustering methods in segmenting health insurance customer data. The data includes age, BMI, lifestyle, and healthcare costs of customers in the United States. Multicollinearity tests showed no multicollinearity issues, allowing all variables to be used in the clustering analysis. The optimal number of clusters was determined using the silhouette coefficient index, indicating that two clusters were the best choice for both methods. Based on the evaluation using the Davies-Bouldin Index (DBI), the k-median method proved superior to FCM by producing more compact and well-separated clusters. K-median clustering results also displayed clearer separation based on categorical variables such as lifestyle. This superiority demonstrates that the k-median method provides more relevant cluster characteristics with significant differences in age, BMI, smoking habits, and healthcare. |
| URI: | http://repository.ipb.ac.id/handle/123456789/160218 |
| Appears in Collections: | UT - Actuaria |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_G5402201030_73f3c30532af4b2da9ea9e75e7f3defb.pdf | Cover | 462.22 kB | Adobe PDF | View/Open |
| fulltext_G5402201030_d3b5584b87b542c2bfe52a438e306d35.pdf Restricted Access | Fulltext | 1.96 MB | Adobe PDF | View/Open |
| lampiran_G5402201030_c7261d9205974f97879619fa0f5c3dfc.pdf Restricted Access | Lampiran | 387.67 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.