In: Statistics and Probability
Barbara Lynch is the product manager for a line of skiwear produced by HeathCo Industries and privately branded for sale under several different names, including Northern Slopes and Jacque Monri. A new part of Ms. Lynch's job is to provide a quarterly forecast of sales for the northern United States, a region composed of 27 states stretching from Maine to Washington. A 10-year sales history is shown:
Period | Sales |
Mar-07 | 72,962 |
Jun-07 | 81,921 |
Sep-07 | 97,729 |
Dec-07 | 142,161 |
Mar-08 | 145,592 |
Jun-08 | 117,129 |
Sep-08 | 114,159 |
Dec-08 | 151,402 |
Mar-09 | 153,907 |
Jun-09 | 100,144 |
Sep-09 | 123,242 |
Dec-09 | 128,497 |
Mar-10 | 176,076 |
Jun-10 | 180,440 |
Sep-10 | 162,665 |
Dec-10 | 220,818 |
Mar-11 | 202,415 |
Jun-11 | 211,780 |
Sep-11 | 163,710 |
Dec-11 | 200,135 |
Mar-12 | 174,200 |
Jun-12 | 182,556 |
Sep-12 | 198,990 |
Dec-12 | 243,700 |
Mar-13 | 253,142 |
Jun-13 | 218,755 |
Sep-13 | 225,422 |
Dec-13 | 253,653 |
Mar-14 | 257,156 |
Jun-14 | 202,568 |
Sep-14 | 224,482 |
Dec-14 | 229,879 |
Mar-15 | 289,321 |
Jun-15 | 266,095 |
Sep-15 | 262,938 |
Dec-15 | 322,052 |
Mar-16 | 313,769 |
Jun-16 | 315,011 |
Sep-16 | 264,939 |
Dec-16 |
301,479 |
a.) Because Ms. Lynch has so many other job responsibilities, she has hired you to help with the forecasting effort. First, she would like you to prepare a time-series plot of the data and to write her a memo indicating what the plot appears to show and whether it seems likely that a simple linear trend would be useful in preparing forecasts.
b.) In addition to plotting the data over time, you should estimate the least-squares trend line in the form:
SALES=a+b(TIME)
Set TIME=1 for 2007Q1 through TIME=40 for 2016Q4. Write the trend equation:
SALES=_________+/-__________(TIME)
(Circle + or - as appropriate)
c.) Do your regression results indicate to you that there is a significant trend to the data? Explain why or why not.
d.) On the basis of your results, prepare a forecast for the four quarters of 2017.
e.) A year later, Barbara gives you a call and tells you that the actual sales for the four quarters of 2017 were: Q1=334,271, Q2=328,982, Q3=317,921, and Q4=350,118. How accurate was your model? What was the mean absolute percentage error (MAPE)?
Result:
a.) Because Ms. Lynch has so many other job responsibilities, she has hired you to help with the forecasting effort. First, she would like you to prepare a time-series plot of the data and to write her a memo indicating what the plot appears to show and whether it seems likely that a simple linear trend would be useful in preparing forecasts.
The plot shows simple linear trend would be useful in preparing forecasts.
b.) In addition to plotting the data over time, you should estimate the least-squares trend line in the form:
SALES=a+b(TIME)
Set TIME=1 for 2007Q1 through TIME=40 for 2016Q4. Write the trend equation:
SALES= 88741.012 + 5362.623 (TIME)
(Circle + or - as appropriate)
Excel Addon Megastat used.
Menu used: correlation/Regression ---- Regression Analysis
Regression Analysis |
|||||||
r² |
0.866 |
n |
40 |
||||
r |
0.931 |
k |
1 |
||||
Std. Error of Estimate |
24961.564 |
Dep. Var. |
Sales |
||||
Regression output |
confidence interval |
||||||
variables |
coefficients |
std. error |
t (df=38) |
p-value |
95% lower |
95% upper |
|
Intercept |
a = |
88,741.012 |
8,043.906 |
11.032 |
0.0000 |
72,456.975 |
105,025.048 |
time |
b = |
5,362.623 |
341.907 |
15.684 |
0.0000 |
4,670.468 |
6,054.777 |
ANOVA table |
|||||||
Source |
SS |
df |
MS |
F |
p-value |
||
Regression |
153,278,654,180.124 |
1 |
153,278,654,180.124 |
246.00 |
0.0000 |
||
Residual |
23,677,027,738.851 |
38 |
623,079,677.338 |
||||
Total |
176,955,681,918.975 |
39 |
|||||
c.) Do your regression results indicate to you that there is a significant trend to the data? Explain why or why not.
To test significance of the model, Calculated F=246.00, P=0.0000 which is < 0.05 level of significance. Ho is rejected. There is a significant linear trend in the data.
d.) On the basis of your results, prepare a forecast for the four quarters of 2017.
Predicted values for: Sales |
|
time |
Predicted |
41 |
308,608.54 |
42 |
313,971.16 |
43 |
319,333.78 |
44 |
324,696.41 |
e.) A year later, Barbara gives you a call and tells you that the actual sales for the four quarters of 2017 were: Q1=334,271, Q2=328,982, Q3=317,921, and Q4=350,118. How accurate was your model? What was the mean absolute percentage error (MAPE)?
time |
observed |
Predicted |
absolute difference |
absolute difference/observed |
41 |
334,271 |
308,608.54 |
25662.4615 |
0.0768 |
42 |
328,982 |
313,971.16 |
15010.8389 |
0.0456 |
43 |
317,921 |
319,333.78 |
1412.7837 |
0.0044 |
44 |
350,118 |
324,696.41 |
25421.5937 |
0.0726 |
mean |
0.0499 |
MAPE = 0.0499*100 = 4.99%