Consider the $$Analytics Exercise: Forecasting Supply Chain Demand$-$Starbucks Corporation$$ PLEASE SHOW ALL ANSWERS AS I NEED TO COMPARE THIS WITH MY WORK.

Part A $(8$ marks$)$

You will develop simple moving average forecast based on the past $5$ weeks and the past $3$ weeks of data and find the MAD, MSE, MAPE, and tracking signal of your forecast. Which forecast, the $5-$week or the $3-$week, is more accurate for each of the distribution centres? Provide a rationale for its better accuracy for that particular warehouse using a text box.

Part B $(8$ marks$)$

Following the question on the next page of the exercise in your textbook, you will develop exponential smoothing forecast based on alpha $=0\mathrm{.}2$ and alpha $=0\mathrm{.}4$ and find the MAD, MSE, MAPE, and tracking signal of your forecast. Which forecast, alpha$=0\mathrm{.}2$ or alpha $=0\mathrm{.}4,$ is more accurate for each of the distribution centres? Which value of alpha emphasizes more on the past data?

Part C $(8$ marks$)$

Starbucks is planning to consolidate the five distribution centres into one with the aggregated demand represented by the total row. Perform forecasts for the $3-$week moving average, $5-$week moving average, $0\mathrm{.}2$ alpha exponential smoothing, and $0\mathrm{.}4$ alpha exponential smoothing on the aggregated total demand. After comparing the results and accuracy measures, provide a rationale on which of these four forecasting methods you would recommend to Starbucks.

Part D $(8$ marks$)$

Starbucks also wants you to develop a linear regression model based on the total demand data for Week $-1$ to $-5.$ Use this model to predict the total aggregated demand for Week $1$ to Week $13,$ and find the forecast accuracy in terms of MAD, MSE, and MAPE. Does linear regression outperform your suggested method in Part C of this question in terms of accuracy measures?

Part E $(4$ marks$)$

Discuss any two pros and two cons of aggregating demand of these five distribution centres.

Part F $(4$ marks$)$

One may develop a new forecast based on the past actual sales or based on the past forecasts. Discuss one benefit that accrues from each approach