Abstract
Insurance provides future protection against setbacks, the significance of insurance in fostering financial
inclusion and driving economic growth cannot be overstated. However, the growth of the insurance
sector has been hindered by low uptake with companies traditionally using actuarial formulas.
Therefore, this paper compared the use of Artificial Neural Networks (ANN) and Autoregressive
Integrated Moving Average (ARIMA) models in predicting Nigerian insurance. The research begins by
subjecting the data to various statistical tests and graphical representations to assess its stationarity and
characteristics. For ARIMA modeling, the study identifies the appropriate model through information
criteria and subsequently estimates, checks, and forecasts the chosen model. Similarly, an ANN model
was constructed, evaluated, and rigorously compared to the ARIMA model in terms of its forecasting
accuracy. The ARIMA (0,2,3) model and the ANN (5-10-1) model were selected based on their
performance metrics. This is evident from the significantly lower forecasting errors across various
evaluation metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean
Absolute Error (MAE) for both in-sample and out-sample forecasts. Results showed that ANNs
outperformed ARIMA, demonstrating lower errors and superior overall performance. The findings
advocate for the adoption of advanced neural network approaches by insurance companies to enhance
their forecasting capabilities. This, in turn, equips insurers with more accurate predictions, enabling
them to make well-informed decisions about future trends within the insurance sector.

Key words: Artificial Neural Networks (ANN); ARIMA; Time Series; prediction; model comparison; Insurance