Year	Month	Sales (in millions)
2017	1	30.2
2017	2	29.8
2017	3	30.8
2017	4	31.5
2017	5	31.9
2017	6	32.6
2017	7	33.8
2017	8	35.9
2017	9	33.2
2017	10	31.5
2017	11	30.9
2017	12	38.8
2018	1	31.1
2018	2	30.5
2018	3	31.5
2018	4	32.5
2018	5	32.7
2018	6	33.2
2018	7	34.5
2018	8	36.5
2018	9	34
2018	10	32.2
2018	11	31.9
2018	12	39.7
2019	1	31.6
2019	2	31.5
2019	3	32.2
2019	4	33.4
2019	5	33.3
2019	6	33.8
2019	7	35.4
2019	8	37.4
2019	9	34.7
2019	10	32.9
2019	11	32.8
2019	12	40.5
2020	1	32.1
2020	2	32.5
2020	3	33
2020	4	34.1
2020	5	33.9
2020	6	34.7
2020	7	36.2
2020	8	38.2
2020	9	35.7
2020	10	33.4
2020	11	33.5
2020	12	41.3

%) Given in the columns A to C are the demands for cell phones from 2017-2020. You are now at the beginning of 2020 and trying to build a forecast model to predict the demand of 2020. Use the records of 2017-2019 to build your model and then estimate the data of 2020 based on the model you have developed.

(a) Before you proceed, graph the past 3 years (2017-2019) data and analyze the visual output.

(b) Set up a 4-period moving average forecast to predict the demands for years 2017-2020. What is the MAD for the forecasts of year 2020?

(c) Set up a 4-period moving average forecast with seasonality (monthly) for years 2017-2020. What is the MAD for the forecasts of the years 2020?

(d) Set up (a) exponential smoothing model (initial forecast = 30 and alpha = 0.5) to predict the demands for year 2017-2020. What is the MAD for the forecasts of year 2020?

(e) Following (d), using what-if analysis, what will be the best value of alpha (one decimal point) so that the MAD for the forecast of year 2020 will be minimized?