Financial and Tax Data Management and Departmental Budget Prediction Method Integrating ARIMA Model

Abstract

Aiming at the problems of low timeliness and inaccurate budgets in the financial sector, the study proposes to apply the time series model to financial and tax data management and departmental budget prediction to realize scientific and accurate prediction of the departmental budget. The method employs a series of tests to ascertain the stationarity of the time series. This is followed by the identification, ordination, and testing of the model to ensure that it will extract all of the useful information from the data. Finally, the constructed time series model is used for the predicting of financial departmental budget data. The experimental results indicated that the departmental budget data of a province from 2015 to 2022 is a non-stationary series, which is converted to a stationary series after the smoothing process. The parameters of the time series model were determined by combining auto correlation function, partial auto correlation function and Akaike information criterion. The determined time series model was used to predict the departmental budget and the average error between the TV and PV of the departmental budget for the years 2020 to 2022 were 7.3%, 7.4% and 12.1%, respectively. Comparing with generalized autoregressive conditional heteroskedasticity model, the prediction precision of the research method in dynamic prediction increased by 26.6%, and the prediction precision of the research method in static prediction was consistent. Therefore, the prediction precision of the research prediction method is high and the error rate is low, which can effectively realize departmental budget prediction analysis and is of practical significance in guiding financial practice.