In: Statistics and Probability
We consider data in the following table, summarizing sales of a product (in thousands). For each of the questions, justify your answer.
Year |
Quarter |
Sales |
2015 |
1 2 3 4 |
4.95 4.25 6.15 6.65 |
2016 |
1 2 3 4 |
5.85 5.35 6.95 7.55 |
2017 |
1 2 3 4 |
6.15 5.75 7.65 7.95 |
(3.1) Assuming the given time-series shows evidence of seasonality, Determine the estimate sales values. (5)
Here we will be considering the average method to obtain the estimate sales values,
To estimate the seasonal relatives, we are going to do it by averaging the demands each period, and dividing by the overall average.
Periods | 2015 | 2016 | 2017 | Average |
Quarter 1 | 4.95 | 5.85 | 6.15 | 5.65 |
Quarter 2 | 4.25 | 5.35 | 5.75 | 5.1167 |
Quarter 3 | 6.15 | 6.95 | 7.65 | 6.9167 |
Quarter 4 | 6.65 | 7.55 | 7.95 | 7.3833 |
OVERALL AVERAGE | 6.267 |
So we have found what the average demand is for Quarter 1 of a year, for Quarter 2, etc. If we divide these averages by the overall average, we get the following seasonal indices:
Periods | Quarter Average | Over-All Average | Seasonality Index |
Quarter 1 | 5.65 | 6.267 | = (5.65/6.267) = 0.9016 |
Quarter 2 | 5.1167 | 6.267 | = (5.1167/6.267) = 0.8165 |
Quarter 3 | 6.9167 | 6.267 | = (6.9167/6.267) = 1.1037 |
Quarter 4 | 7.3833 | 6.267 | = (7.3833/6.267) = 1.1782 |
Now, to estimate the sales, we will de-seasonalize the individual sales numbers to see how sales per period go up or down.
The de-seasonalizing method uses the seasonal relatives we already created to try to get a picture of how much we’ve been growing over time. What is deseasonalizing? After we make a straight-line forecast of the future, we are going to multiply it by the seasonal indices to get a seasonalized forecast of the future. Deseasonalizing is basically the opposite: we are going to take the actual, seasonal data, and divide it by the seasonal factors to get something that looks more like a straight line. Then we are going to do a linear regression through this line. The deseasonalized demands should be a lot more like a straight line than the original data is, that is, it should generally show a more consistent 3 growth rate than we see with seasonality in it. When we do a linear regression through these deseasonalized points, the linear regression should give us a pretty good fit through the points.
Quarters | Seasonal Index | Deseasonal Index | Final Value |
Quarter 1 - 2015 | 4.95 | 0.9016 | = (4.95/0.9016) = 5.49 |
Quarter 2 - 2015 | 4.25 | 0.8165 | = (4.25/0.8165) = 5.21 |
Quarter 3 - 2015 | 6.15 | 1.1037 | = (6.15/1.1037) = 5.57 |
Quarter 4 - 2015 | 6.65 | 1.1782 | = (6.65/1.1782) = 5.64 |
Quarter 1 - 2016 | 5.85 | 0.9016 | = (5.85/0.9016) = 6.49 |
Quarter 2 - 2016 | 5.35 | 0.8165 | = (5.35/0.8165) = 6.55 |
Quarter 3 - 2016 | 6.95 | 1.1037 | = (6.95/1.1037) = 6.30 |
Quarter 4 - 2016 | 7.55 | 1.1782 | = (7.55/1.1782) = 6.41 |
Quarter 1 - 2017 | 6.15 | 0.9016 | = (6.15/0.9016) = 6.82 |
Quarter 2 - 2017 | 5.75 | 0.8165 | = (5.75/0.8165) = 7.04 |
Quarter 3 - 2017 | 7.65 | 1.1037 | = (7.65/1.1037) = 6.93 |
Quarter 4 - 2017 | 7.95 | 1.1782 | = (7.95/1.1782) = 6.75 |
Now, we do a linear regression through these deseasonalized numbers :
We get an intercept of 5.24 and a slope of 0.16, and an R2 value of 0.82.
Thus, multiplying the linear forecast we made by the seasonal factors to get a seasonalized estimated forecasts are:
Time Period | Estimated Trend | Estimated Forecast |
1 | 5.400417291 | 4.86901623 |
2 | 5.557916104 | 4.538038499 |
3 | 5.715414918 | 6.308103445 |
4 | 5.872913731 | 6.919466958 |
5 | 6.030412545 | 5.43701995 |
6 | 6.187911358 | 5.052429624 |
7 | 6.345410172 | 7.003429207 |
8 | 6.502908985 | 7.661727366 |
9 | 6.660407799 | 6.005023671 |
10 | 6.817906612 | 5.566820749 |
11 | 6.975405426 | 7.698754968 |
12 | 7.132904239 | 8.403987774 |