Abstract: Forecasting of insurance claims is of great concern to insurance industry. In motor insurance, claim payments constitute to a significant portion of insurance’s expenditure, making accurate forecasting an essential aspect. Traditional models such as Generalized Linear Models which has been widely used in predicting insurance claims often fail to capture seasonality, trends and temporal dependencies in the data leading to inaccurate forecasts. This research applied a SARIMA model to predict motor insurance claims in Kenya. The quarterly motor insurance data from 2017-2024 was obtained from IRA and analyzed through Box-Jenkins Methodology. From the time series plot, it was found that the data exhibit seasonality, with claims paid each quarter increasing continuously from the first quarter to the last in each year. SARIMA (1,1,1) (1,1,0,4) was chosen using Grid Search Optimization since it had the lowest AIC value (320.5). The suitability of this model was also confirmed through model diagnosis. A 2-year forecast graph showed a rising trend in motor insurance claims while still maintaining seasonal fluctuations that aligns well with past data. The future confident intervals widened with time indicating that there is an increase in uncertainty of the forecasts. From the analysis, the study suggests that SARIMA is a better tool for projecting seasonal motor insurance claims in Kenya. Motor insurers will minimize losses that results from inaccurate forecast by utilizing this model.

Keywords: SARIMA, Claims, Motor Insurance.

| DOI: 10.17148/IARJSET.2025.12611