|
|
 |
| |
|
|
|
On Forecast Performance of Autoregressive Integrated Moving Average Model in Nigerian Stock Exchange |
|
|
|
PP: 33-39 |
|
|
|
doi:10.18576/jsapl/080104
|
|
|
|
Author(s) |
|
|
|
K. K. Ajibodu,
O. K. Agunloye,
|
|
|
|
Abstract |
|
|
| This paper examines forecast performance of two classes of ARIMA models namely ARIMA (1, d, q) and ARIMA (p, d, 1) models in forecasting all shares index of Nigerian stock exchange. The stationarity of all shares index of Nigerian stock exchange data was examined using augmented Dickey-Fuller test and the data becomes stationary after second differencing showing that data is integrated of order 2. The choice of autoregressive and moving average order is another important specification decision that requires serious attention to avoid the problem of over-fitting or under-fitting of model order. Under model identification, ten ARIMA models were fitted comprising five different ARIMA specifications from each class of ARIMA models under consideration. Using standard information criteria such as Akaike information criterion and Bayesian information criterion, we selected two optimal models comprising ARIMA (1, 2, 5) model from the class of ARIMA (1, d, q) models and ARIMA (5, 2, 1) model from the class of ARIMA (p, d, 1) models. Using the forecast accuracy measures such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Mean Square Error (MSE), we demonstated that ARIMA (1, 2, 5) model has the best in-sample and out-of-sample forecast performance than ARIMA (5, 2, 1) model.
|
|
|
|
|
|
|
|
