A study on predicting the number of accesses to information systems using the Prophet model

Even for the Construction Technology Information System, which receives support from the government's informatization budget, accurate performance target setting is crucial for assessing performance. This study defined the number of system accesses as a performance indicator and aimed to predict the number of accesses over the past two years. To achieve this, accesses were collected from January 2025 to December 2026 and determined to be non-stationary trend-type time series data. The ARIMA, GARCH, and Prophet models, which are representative models used in econometrics to characterize time series data, were selected. These models were applied to obtain forecasts and volatility of accesses between 2025 and 2026. The forecasts were compared between the ARIMA and Prophet models. The Prophet model was found to be more statistically significant than the ARIMA model for accesses. Furthermore, the GARCH model was applied to analyze the volatility characteristics of the forecasts. We found that trend-type time series data, such as the number of connections, which increase and decrease at a constant level, can estimate volatility, but are not significant in reflecting the intensity and persistence of volatility.