An ARIMA based model for forecasting the patient number of epidemic disease
Yanchun Pan, Mingxia Zhang, Zhimin Chen, Ming Zhou, Zuoyao Zhang
Abstract
Yanchun Pan, Mingxia Zhang, Zhimin Chen, Ming Zhou, Zuoyao Zhang
Abstract
Forecasting the number of epidemic disease is very important for CDC (center for disease control and prevention). To improve the forecast accuracy, an ARIMA (autoregressive integrated moving average) based model is proposed in this paper. First, autocorrelation (AC) and partial autocorrelation (PAC) analysis are introduced to establish a stationary time series, where the autocorrelation order, moving average order and difference order are estimated. Secondly, least square s method (LS) is employed to estimate the parameters of the prediction model. Finally, the real data between Jan. and Aug. 2014 coming from a CDC are fed into the proposed model and the forecast accuracy obtained is 92.1%, which significantly outperforms the simple moving average method currently used in the CDC.
OpenAlex reports 30 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Forecasting the number of epidemic disease is very important for CDC (center for disease control and prevention). To improve the forecast accuracy, an ARIMA (autoregressive integrated moving average) based model is proposed in this paper. First, autocorrelation (AC) and partial autocorrelation (PAC) analysis are introduced to establish a stationary time series, where the autocorrelation order, moving average order and difference order are estimated. Secondly, least square s method (LS) is employed to estimate the parameters of the prediction model. Finally, the real data between Jan. and Aug. 2014 coming from a CDC are fed into the proposed model and the forecast accuracy obtained is 92.1%, which significantly outperforms the simple moving average method currently used in the CDC.
Key concepts: Autoregressive integrated moving average, Autocorrelation, Partial autocorrelation function, Moving-average model, Autoregressive model, Moving average, Time series, Computer science