Implementasi Data Mining dalam Klasterisasi Prevalensi Stunting di Kabupaten Kulon Progo
Hasna' Shofiyyah, Dr. Nur Rokhman, S.Si., M.Kom.
2026 | Tugas Akhir | D4 MANAJEMEN INFORMASI KESEHATAN
Latar Belakang: Stunting pada balita masih menjadi isu yang serius, dengan prevalensi nasional 19,8% pada 2024 dan belum memenuhi target 14,2% di tahun 2029. Di Kabupaten Kulon Progo, angka stunting menunjukkan tren yang fluktuatif dan belum sepenuhnya terkendali. Analisis indikator secara terpisah belum mampu menggambarkan pola antarwilayah secara komprehensif, sehingga diperlukan pendekatan yang mampu mengintegrasikan berbagai indikator dan menyajikan informasi secara lebih mudah dipahami untuk mendukung pengambilan keputusan berbasis data.
Tujuan: Mengadaptasi dan mengimplementasikan kerangka kerja data mining untuk mengelompokkan objek puskesmas-tahun di Kabupaten Kulon Progo berdasarkan indikator stunting dan cakupan layanan kesehatan, serta menyajikan hasilnya dalam bentuk dashboard.
Metode: Penelitian ini termasuk kuantitatif dengan pendekatan cross-sectional menggunakan data sekunder indikator stunting dan cakupan layanan kesehatan dari laporan tahunan Dinas Kesehatan Kulon Progo periode 2019-2025. Analisis dilakukan dengan hierarchical clustering Ward’s method berbasis Python. Hasil disajikan melalui dashboard Looker Studio.
Hasil: Kerangka kerja data mining diadaptasi dari CRISP-DM dalam enam tahapan, mulai dari pengumpulan data hingga penyajian dashboard. Klasterisasi terhadap 147 objek puskesmas-tahun menghasilkan empat klaster, masing-masing mencerminkan profil puskesmas yang berbeda berdasarkan tingkat stunting dan cakupan layanan kesehatan pada periode tertentu. Dashboard interaktif yang dikembangkan dilengkapi fitur filter untuk mendukung eksplorasi hasil.
Kesimpulan: Kerangka kerja data mining yang diadaptasi dari CRISP-DM berhasil diimplementasikan untuk mengelompokkan puskesmas-tahun di Kabupaten Kulon Progo menjadi empat klaster berdasarkan indikator stunting dan cakupan layanan kesehatan. Dashboard interaktif yang dikembangkan dapat memfasilitasi eksplorasi profil puskesmas secara informatif.
Background: Stunting among children under-five remains a serious issue, with a national prevalence of 19.8% in 2024, which has not yet met the 14.2% target set for 2029. In Kulon Progo Regency, stunting prevalence shows fluctuating trends and has not yet been fully brought under control. Analyzing indicators separately has been insufficient to comprehensively capture interregional patterns, highlighting the need for an approach that integrates multiple indicators and presents information in a more accessible manner to support data-driven decision-making.
Objective: Adapt and implement a data mining framework to cluster primary health center-year observations in Kulon Progo Regency based on stunting indicators and health service coverage, and present the results in the form of a dashboard.
Methods: This study used a quantitative cross-sectional design using secondary data on stunting indicators and health service coverage from the annual reports of the Kulon Progo Health Office for the period 2019-2025. The analysis was conducted using Ward's hierarchical clustering method in Python. The results are presented via a Looker Studio dashboard.
Results: The data mining workflow was adapted from CRISP-DM and consists of six phases, ranging from data collection to dashboard presentation. Clustering of 147 primary health center-year observations yielded four clusters, each reflecting a different primary health center profile based on stunting rates and health service coverage achievements during a specific period. The interactive dashboard developed includes filter features to support the exploration of results.
Conclusion: A data mining workflow adapted from CRISP-DM was successfully implemented to group primary health center-year observations in Kulon Progo Regency into four clusters based on stunting indicators and health service coverage. The interactive dashboard developed can facilitate the informative exploration of primary health center profiles.
Kata Kunci : Data Mining, Hierarchical Clustering, Stunting, Puskesmas, Dashboard, Data Mining, Hierarchical Clustering, Stunting, Primary Health Centers, Dashboard