Right, so I've highlighted above the important parts of this question. First we need to use the regression line you worked out in part c - according to the answers, it was:
But first we need to put in
the third quarter of 2004, which can be calculated like this:
Where 1 is the number of years that have passed, 4 is the number of seasons (well, quarters) in that year and the +3 is to represent the third quarter of the next year. So our quarter 7. This could also be found by counting it out on the table or graph, but... I like formulas, so nyuh.
So now we can substitute that into our regression equation to make a prediction like everything else in the core module:
(rounded to 2 decimal places)
But wait a second here, that prediction is wayyyy off. The actual value is 128. We're not finished yet, we need to multiply this by the appropriate seasonal index in order to get a deseasonalised value. I'm going to work out the seasonal index because a), I couldn't see it in the answers, and b) people often make a mistake here and that might be your issue
.
(I am, however, skipping through the seasonal averages here because I'm too lazy to type that out. Unless you want it, in which case just let me know)
Ignore how that's messy, on a question like this I'd usually just throw that straight into my calculator. There's two values on top because we're working out a kind of average between the two appropriate seasonal indices so we can make a (slightly?) more accurate prediction. So our seasonal index is:
So like I said before, we just multiply this by our predicted value to deseasonalise it:
This is still not quite our answer, because the question states that all values are in thousands of dollars. All we need to do is multiply by 1000, so the sales in dollars for the third quarter of 2004 is (was? It's been ten years.
) $128,174. My answer is 4 dollars off of the one you showed, because my rounding was a teensy bit inconsistent (hey, I blame the question for not stating how many decimal places!), but the method is the same.