ARIMA modeling - Assignment Example

In other words, the data for modeling the revenues of Costco Company may probably need differentiation through it is not certain that there is need for differentiation. In addition, the fact that the first six lags fall outside the confidence area shows that the data is non-stationary, that is, there is trend in the revenue function of Costco Company. Even though there is a trend (non-stationary data) it does not necessarily mean that the data should be transformed or differentiated. This decision can only be reached if specific lags such as 12, 24, and 36 are verified in respect to their expected values.

That is when the decision to differentiate the data to remove seasonality and trend will be arrived at. In as much as the aspects of being non-stationary and having a trend have been removed through AR(1) as depicted in the ACF graph, the data looks much better though with two positive spikes at lag 1 and lag 3,. The lags 1 and 3 shows that the data is still non-stationary and there is trend in the data. There is need to remove the non-stationary and trend aspects of the data for efficient and effective modeling of the problem.

The indication is that the data does not have a mean or constant variance. In order to do this, there is need to further differentiate the data by taking the 3rd different of y since this will remove the seasonality in the data. The above graph shows that there are specific lags that lie outside the confidence area. This means that the data is not stationary. It is important to find the ACF for the other differences such as 1st, 4th, 5th, 6th, and 7th for the purposes of removing aspects of seasonality in the data.

These lags exist outsides the confidence level as depicted by the ACF graph. The coefficients are SAR and SMA due to the seasonality present in the data. The p values of both coefficients are below .05. The MS is 466718 for the model.