The Society of Convergence Knowledge Transactions ISSN:2287-8920(Print) 융복합지식학회논문지

In this study, a KOSDAQ index prediction model using the transition function model is presented using 12-year time series data. In order to apply the transition function model, the producer price index and KOSDAQ index were first transformed into stationary time series by square root transformation, first order differencing, and seasonal differencing, and then the unit root test was performed to confirm that it is stationary time series. In order to determine the inclusion of a constant term, a T-test on the mean was performed. The pre-whitening process was performed on the producer price index, which was an input time series, to estimate the pre-whitening parameter of the KOSDAQ index, and the transition function model was determined by estimating the cross-correlation function and impact response weight between the producer price index and the KOSDAQ index. The autocorrelation analysis of the transition function model was performed to estimate the noise time series model, and the KOSDAQ index prediction model was presented by combining the transition function model with the noise model. The goodness of fit of the estimated prediction model was evaluated by the white noise probability process using the mean test on the residuals and the Portmanteau test statistics. The prediction model was confirmed to be fit by the results of the Portmanteau test on the existence of the cross-correlation between the residuals and the producer price index, and subsequently, the KOSDAQ index prediction values and intervals were presented using the transition function model.