The Application Of Model For Combination Forecasting In The Study Of Outpatient Volume At Hengdian Group Hospital In Dongyang

Objective:To accurately predict the trend of future outpatient volume of Hengdian Group Hospital of Dongyang, a comparison between individual and the combination of ARIMA-time-series and the GM(1,1) models was made to select the best applicable model.Methods:The monthly outpatient volume between January 2003 and December 2011 was collected. Based on the monthly outpatient data from January 2003-June 2011, ARIMA and GM(1,1) models were first used as fitting methods. The optimal weighting method based on the combination of above tow methods was then established. All three models were examined for their accuracy to predict the outpatient volume from July to December of 2011.Results:1. Based on the volume between Jan.2003 and Jun.2006, the best fitting equation for ARIMA-time-series is Z(T+L)= -0.0024+0.15Z(T+L-4) +e(T+L)-0.24e(T+L-4), residual sequence is white noise with degree of fitting C=101.82%, AIC-424.79. As for fitted values, Mean Square Error (MSE) is 21.52; Mean Absolute Error is (MAE) 161.62; Mean Absolute Percentage Error (MAPE) is 1.16%; Mean Square Percentage Error (MSPE) is 1.4%. When used to predict outpatient volume between Jul. to Dec.2011, the valuves of MSE, MAE, MAPE and MSPE are 803.97, 1779.26,9.87%,3.90% respectively.2. When GM(1,1) was applied to fit the outpatient volume of Hengdian Group Hospital from Apr.2003 to Jun.2011, the fitting equation is x(t+l)= 1097615.86e-0.0083t-1088837.86 with degree of fitting 73.06%, MSE 198.90, MAE 1432.55, MAPE10.30%, MSPE1.43%. When it was used to predict outpatient volume of Jul. to Dec.2011, MSE, MAE, MAPE, and MSPE are 893.22,1770.48,9.6%, 5.03% respectively.3. With combinational model, the equation is yt=0.9840y11+0.0160y12. The degree of fitting is 100.47%, MSE 21.28, MAE 158.52, MAPE1.15%, and MSPE 0.15%. When used to predict the outpatient volume between Jul. and Dec.2011, MSE is 798.55, MAE 1768.40, MAPE 8.83%, MSPE 3.88%.Conclusion: Based on data of Hengdian Group Hospital of Dongyang, combinational model is better than ARIMA-time-series and GM(1,1) models in predicting outpatient volume.