Question

3) (20 pts) If you used a two-year moving average on the data below, how many iPhone’s would you expect to sell in 2019? What is the MAD for this forecasting method?

Year    iPhone sales (# of phones)

2014    125,000,000

2015    150,000,000

2016    169,000,000

2017    231,000,000    

2018    212,000,000

b) If you used a 3-year weighted moving average, how many iPhones would you expect to sell in 2019 if you assign weights of 20%, 30%, and 50% with the higher weights assigned to more recent forecasts?   

c) Between 2-year moving average and 3-year weighted moving average, which method would you choose, and why?

d) There are other forecasting methods different from (a) and (b) above.  Describe one, along with an advantage and disadvantage of using that method here.

Answer 1

Answer a=

Year	Sales ( in 000,000)	2 year moving average ( in 000,000)	2 year moving average( in 000,000)
2014	125
2015	150
2016	169	(150+125)/2	137.5
2017	231	(169+150)/2	159.5
2018	212	(231+169)/2	200
2019		(212+231)/2	221.5

Forecast for 2019=221500000

Year	Sales	2 year moving average	Deviation
2014	125000000
2015	150000000
2016	169000000	137500000	31500000
2017	231000000	159500000	71500000
2018	212000000	200000000	12000000
		Sum	115000000

MAD=115000000/3=38333333

Answer b=

Year	Sales ( in 000,000)	3-year weighted moving average ( In 000,000)	3-year weighted moving average ( In 000,000)
2014	125
2015	150
2016	169
2017	231	(125*0.2+150*0.3+169*0.5)/(0.2+0.3+0.5)	154.5
2018	212	(150*0.2+169*0.3+231*0.5)/(0.2+0.3+0.5)	196.2
2019		(169*0.2+231*0.3+212*0.5)/(0.2+0.3+0.5)	209.1

Sales forecast for 2017=154000000

Sales forecast for 2018=196200000

Sales forecast for 2019=209100000

Answer C=

Year	Sales	3-year weighted moving average	Deviation
2017	231000000	154500000	76500000
2018	212000000	196200000	15800000
		Sum	92300000

MAD for 3 year weighted moving average = 92300000/3 =46150000

I will select 2 year moving average as it has lesser MAD

Answer d= We can also use exponential smoothing method that is used for smoothing time series data

Advantage=

It is simple to use

Only limited data is needed

Disadvantage=

There is lagging behind the actual data

It doe snot include any trend or seasonality