Tanah longsor merupakan bencana hidrometeorologi yang kerap terjadi di Kabupaten Kerinci dan Kota Sungai Penuh, Provinsi Jambi. Penelitian ini bertujuan untuk memodelkan kerentanan tanah longsor secara spasial menggunakan dua pendekatan statistik, yaitu regresi logistik dan frekuensi rasio, serta mengevaluasi performa akurasi masing-masing model. Sebanyak 51 titik longsor hasil survei lapangan digunakan, dengan pembagian 80% sebagai data pelatihan dan 20% untuk validasi. Sebelas variabel fisik digunakan sebagai faktor prediktor: kemiringan lereng, elevasi, aspek lereng, bentuk lereng, jarak dari jalan, jarak dari sungai, jarak dari patahan, Topographic Wetness Index (TWI), Stream Power Index (SPI), Stream Transport Index (STI), dan NDVI. Hasil analisis menunjukkan bahwa distribusi kerentanan tidak bersifat acak, melainkan terstruktur mengikuti kondisi morfologi dan aktivitas antropogenik. Model frekuensi rasio mengidentifikasi zona kerentanan sangat tinggi seluas 20,31% (77.389,45 ha), sedangkan regresi logistik seluas 18,34% (69.868,96 ha). Validasi menggunakan Area Under Curve (AUC) dari kurva ROC menunjukkan bahwa regresi logistik memiliki tingkat akurasi yang lebih tinggi (AUC = 0,9351) dibanding frekuensi rasio (AUC = 0,7970). Dengan mempertimbangkan interaksi antar variabel, regresi logistik terbukti lebih unggul dalam memodelkan kerentanan tanah longsor secara spasial di wilayah studi.

---

Landslides are a common hydrometeorological hazard in Kerinci Regency and Sungai Penuh City, Jambi Province, Indonesia. This study aims to spatially model landslide susceptibility using two statistical approaches: logistic regression and frequency ratio, and to evaluate the predictive performance of each method. A total of 51 landslide points obtained from field surveys were used, with 80% allocated for training and 20% for validation. Eleven physical variables were employed as predictor factors: slope, elevation, aspect, curvature, distance to roads, distance to rivers, distance to faults, Topographic Wetness Index (TWI), Stream Power Index (SPI), Stream Transport Index (STI), and Normalized Difference Vegetation Index (NDVI). The results indicate that the spatial distribution of susceptibility is not random but structured, following morphological conditions and anthropogenic activities. The frequency ratio model identified very high susceptibility zones covering 20.31% (77,389.45 ha), while logistic regression identified 18.34% (69,868.96 ha). Validation using the Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC) curve showed that logistic regression yielded a higher accuracy (AUC = 0.9351) compared to the frequency ratio (AUC = 0.7970). By accounting for variable interactions, logistic regression proved to be more effective in spatially modeling landslide susceptibility within the study area.