Question 1

The following data (stored in file: STRIKES_data.MTJ) shows the number of strikes in the U.S. from 1951 to 1980:

1. Obtain a time series plot for the data using Minitab.
2. Perform a SLR of Strikes vs Time. Discuss the suggestion that this is a good model for predicting the number of strike.
3. Obtain forecasts for the number of strikes for the years 1981 through 1985
4. Produce the graph of standardized residuals vs Time. Comment on the graph with regard to the assumptions of the regression model.
5. Find the value of the Durbin-Watson statistic and discuss its relevance to this model.
6. Perform an autoregressive model with a 1 period lag. Use a SLR model ( not the ARIMA feature of Minitab).
7. Discuss this model and compare it with the regression of Strikes vs Time. Which model do you recommend? Explain.
8. Using the AR(1) model, forecast strikes for the years 1980 – 1985 inclusive. Compare these forecasts with the one obtained by the regression of Strikes on Time. Which do you think is more realistic. Explain.
9. Using the Time Series feature of Minitab, generate the ACF and PACF graphs for the Strike data.
   From the graphs, would you expect an AR model or an MA model to best describe the data? How many lagged terms would you include in your proposed model? Explain.
10. Use the ARIMA feature of Minitab to specify the appropriate model. Check to see if the ACF and PACF of the residuals are consistent with the hypothesis that the residuals should consist of white noise.
    
11. Compare the AR(1) model obtained from running a simple regression of Y _t against Y _t-1 with the corresponding model obtained by using ARIMA. The ARIMA model does not provide a value of R ² _adj . How can you tell that the ARIMA model has good predictive power?

Price: $16.55

Solution: The downloadable solution consists of 10 pages, 655 words and 11 charts.
Deliverable: Word Document