The territory of the Kingdom of Saudi Arabia is relatively safe and does not contain large landslide-prone regions. However, the west coast of the Arabian Peninsula is susceptible to landslide activity. Therefore, monitoring of these regions is urgently needed. The best way to observe the areas susceptible to landslide activity is the application of remote sensing technologies, particularly InSAR. The presented paper analyzes the monitoring results in Jizan province using InSAR observations. The primary goal was to develop a landslide-forecasting model using various data types. The InSAR monitoring data cover observation epochs from 2020 to 2023. Since the monitoring region lacks reliable reflecting surfaces, the displacements were obtained using the SBAS processing algorithm in the open-source Python package Miami INsar Time-series software. Apart from displacements, meteorological parameters and the landslide susceptibility index were used as independent variables for forecasting model creation. The best way to gain the advantages of different data fusion techniques is by employing a machine learning approach. The group method for data handling (GMDH) algorithm was used for the forecasting model simulation. Different GMDH processing strategies were tested, and the optimal one was selected. The Combinatorial GMDH and Generalized Iterative Algorithm GMDH provided the best prediction outcomes. The analysis only showed an acceptable correlation between displacements and precipitations. The correlation between temperature and displacements was positive but statistically insignificant. GMDH algorithms demonstrated high efficiency and flexibility in analyzing such complex data. The simulation results show that points placed in medium- and high-landslide-activity areas have an average settlement velocity of around -7 ± 2 mm/year and an average uplift velocity of around +9 ± 1 mm/year for some regions.