<![CDATA[Indonesia's South Sumatra province generates substantial four-wheeled vehicle registration activity, yet the administrative records that document this activity have rarely been subjected to systematic analytical scrutiny. This study draws on Samsat vehicle registration data from 2025 and the first quarter of 2026 to examine how new vehicle adoption patterns differ across engine displacement (CC) segments. The dataset spanned 15 Excel files: Dataset A covered the full year 2025 (17,867 units), while Dataset B captured Q1 2026 (6,190 units). Because column naming conventions varied across files, a keyword-matching-based preprocessing pipeline was developed to handle integration. The analysis combined the Mann-Kendall Trend Test, Sen's Slope Estimator, STL Decomposition, Kruskal-Wallis Test, and Mean Absolute Percentage Error (MAPE) as an out-of-sample projection benchmark. The Mann-Kendall test returned no statistically significant monotonic trend (p = 0.7105), which indicates that the proportional split between engine displacement categories stayed broadly stable over the observation window. Low-displacement vehicles (≤1,500 cc) held a consistent majority, averaging 66.10% of monthly registrations, throughout the period. STL Decomposition revealed a seasonal component responsible for roughly 50.02% of total data variance; the low-displacement segment peaked in November, while the high-displacement segment peaked in March. Projection accuracy, measured by MAPE, came in at 14.46% for the low-displacement segment (Accurate) and 37.92% for the high-displacement segment (Moderately Accurate). Based on these results, Samsat's administrative data proves to be highly useful for practical applications. Regional governments can utilize this data to improve public service planning and estimate regional own-source revenue (PAD) more accurately]]>
