Part A 

---

First double click the series "sales" and then choose "unit root test" from "view".
   
And then use this test to determine the value of d.
1) Level, intercept
   
2) Level, Trend and intercept
   
3) Level, none
   
4) 1st diff, intercept
   
5) 1st diff, Trend and intercept
   
6) 1st diff, none
   
And then from the ADF test statistics, d = 1.
Now click the "view", "correlogram", enter the lag length correctly.
And here are the results:
   
Now use the Rats program to examine the Q-stat of every combination,
From the results, ARIMA(3,1,10), ARIMA(4,1,10) and ARIMA(2,1,10)
having small Q-stat value. (0912.prg) (0912.out)
 
And now forecast the models and see the results:
e.g. ARIMA(2,1,10)
 
In the "Estimate Equation", type in information like this:
     
And then click the forecast buttom and enter information like this:
     

---