Analisis Clustering Persepsi Santriwati Akhir KMI Terhadap Pengabdian UNIDA Reguler Menggunakan Algoritma K-Means

Abstract

Penelitian ini bertujuan untuk mengelompokkan dan menganalisis persepsi santriwati akhir Kulliyatul Mu’allimat al-Islamiyyah (KMI) terhadap program pengabdian UNIDA Reguler menggunakan algoritma K-Means clustering. Pengelompokan dilakukan untuk mengetahui tingkat penerimaan santriwati secara objektif, efisien, dan transparan sebagai dasar evaluasi serta pengembangan program pengabdian di masa mendatang. Metode penelitian menggunakan kerangka kerja CRISP-DM, meliputi tahapan data understanding, data preparation, modeling, evaluation, dan deployment. Data primer diperoleh dari kuesioner skala Likert lima poin yang diisi oleh 561 responden. Analisis dilakukan melalui pembersihan dan normalisasi data, pemodelan dengan algoritma K-Means, penentuan jumlah klaster optimal menggunakan Elbow Method, dan validasi dengan Silhouette Score. Hasil penelitian menunjukkan jumlah klaster optimal sebanyak tiga, yaitu klaster positif (25%), netral (41%), dan negatif (34%). Klaster positif memiliki motivasi dan sikap tinggi, klaster netral menunjukkan skor sedang dengan aspek sosial lemah, sedangkan klaster negatif rendah di hampir semua variabel, terutama motivasi dan pengalaman. Nilai Silhouette Score sebesar 0,61 menunjukkan bahwa kualitas klasterisasi tergolong baik. Penelitian ini membuktikan bahwa penerapan K-Means clustering efektif dalam memetakan persepsi santriwati secara sistematis dan akurat. Hasilnya memberikan masukan praktis bagi UNIDA untuk memperkuat motivasi, dukungan sosial, serta strategi pembinaan dan pendampingan agar program pengabdian diterima lebih positifThis study aims to categorise and analyse the perceptions of female students graduating from Kulliyatul Mu'allimat al-Islamiyyah (KMI) towards the UNIDA Regular community service programme using the K-Means Clustering algorithm. The categorisation was conducted to determine the level of acceptance among female students in an objective, efficient, and transparent manner as a basis for evaluation and development of future community service programmes. The research method used the CRISP-DM framework, which includes the stages of data understanding, data preparation, modelling, evaluation, and deployment. Primary data was obtained from a five-point Likert scale questionnaire completed by 561 respondents. The analysis was carried out through data cleaning and normalisation, modelling with the K-Means algorithm, determining the optimal number of clusters using the Elbow Method, and validation with the Silhouette Score. The results showed that the optimal number of clusters was three, namely positive (25%), neutral (41%), and negative (34%). The positive cluster had high motivation and attitude, the neutral cluster showed moderate scores with weak social aspects, while the negative cluster was low in almost all variables, especially motivation and experience. The Silhouette Score value of 0.61 indicates that the clustering quality is good. This study proves that the application of K-Means Clustering is effective in mapping female students' perceptions systematically and accurately. The results provide practical input for UNIDA to strengthen motivation, social support, and coaching and mentoring strategies so that the community service programme is received more positively