Question

If i have 50 randomly generated numbers and I want to apply the following forecasting techniques on them; Simple moving average, Exponential smoothing and trend projection with regression, which would be the best technique between them and why will I choose this one?

Week	Demand
1	688
2	681
3	643
4	667
5	642
6	683
7	758
8	791
9	751
10	682
11	609
12	635
13	647
14	796
15	676
16	617
17	704
18	722
19	764
20	726
21	692
22	708
23	674
24	606
25	640
26	785
27	717
28	684
29	665
30	713
31	711
32	666
33	779
34	756
35	691
36	641
37	617
38	729
39	768
40	621
41	771
42	760
43	684
44	683
45	603
46	665
47	720
48	692
49	704
50	647

Answer 1

Note that the numbers are generated randomly i.e. from a uniform distribution. So, it is expected that the resultant series will not have any trend or seasonality. It will only be a random variation or white noise.

Therefore, there is no need to go for regression as there will be no trend.

Among simple exponential smoothing and smile moving average, a simple moving average method will be the best method in this case because we do not have to account for trend or seasonal pattern and the moving average is the simpler among the two.