<![CDATA[This research is conducted to predict the number of visitors to libraries within primary education institutions by employing the Autoregressive Integrated Moving Average (ARIMA) modeling technique. The dataset comprises daily visitor records spanning from January 2023 to December 2024. The forecasting process adopts a time series framework, which includes steps such as data preprocessing, stationarity verification through the Augmented Dickey-Fuller (ADF) test, identification of parameters using Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) plots, and the selection of the optimal model based on statistical significance and performance metrics, particularly the Mean Squared Error (MSE). Out of 35 evaluated ARIMA configurations, the ARIMA(2,0,11) model demonstrated the best performance, achieving the lowest MSE score of 789.08 and exhibiting statistically meaningful parameters. Moreover, the model passed the Ljung-Box diagnostic test, confirming that the residuals behave as white noise.The forecasting results for January 2025 show a stable and realistic trend. Compared to baseline methods such as Naïve Forecast, the ARIMA model demonstrates superior performance by effectively capturing data fluctuations. Therefore, ARIMA(2,0,11) is considered effective and accurate in supporting data-driven library service planning for the future.]]>
Copyrights © 2025
