Mini-project 3: Forecasting

Eli Orchid has designed a new pharmaceutical product, Orchid Relief that improves night sleep. (We will revisit this company in future mini-projects as well). Before mass production of the product, Eli Orchid has market tested Orchid Relief in only Orange County over the past 8 weeks. The weekly demand is recorded. Eli Orchid is now trying to use the sales pattern over the past 8 weeks to predict sales in US for the upcoming few months. An accurate forecast would be much helpful in arrangements for the company’s production processes and design of price promotions over each week. The data is provided in file EliOrchid_OC demand.xlsx.

After reviewing the data, the marketing manager of Eli Orchid believes that there is a repeating pattern in Orchid Relief use likely due to changes in sleeping pattern of customers on different days.

1) How many seasons are in each repeating cycle?

a) 3 b) 5 c) 7 d) 12

2) To make a forecast of day 60, what is the seasonal factor for this day?

a. 90.58% b. 98.02% c. 99.05% d. 100% e. 121.89%

250

300

350

400

450

500

0 5 10 15 20 25 30 35 40 45 50 55 60

Daily Demand

3) Looking at the deseasonalized time series, what is the slope of a linear time trend in the deseasonalized data?

a. 339.29 b. 351.08 c. 4.018× 10−7 d. 0.6143 e. 1.0356

4) Can we conclude at 5% significance level that there is a significant linear time trend in the

deseasonalized time series? a. Yes b. No c. Cannot be determined

5) Using the regression method on the deseasonalized time series, what is a deseasonalized forecast

for day 60? a. 343.43 b. 353.54 c. 387.94 d. 401.42

6) Taking seasonality into account, what is a forecast for day 60?

a. 387.94 b. 351.40 c. 353.54 d. 363.61

After the above forecasts are presented to the Chief Marketing Officer, he stated that a special marketing campaign has been launched on day 53. Therefore, under this new situation, the firm is unable to rely too much on data that is older than 4 days (before the campaign starts). However, he still believes that seasonality exists, and getting reliable seasonal indexes requires much more data than just 4 data points.

Chief Marketing Officer suggested this new approach to make forecasts: Use seasonal indexes and deseasonalized values you have got earlier to deal with seasonality. But to make a deseasonalized forecast, look only at the last 4 deseasonalized data points (for days 53-56) and assume that these 4 data points form a stationary time series. Make a deseasonalized forecast based on these 4 pieces of data with the methods suggested in questions 7 – 10 as you would do for any stationary time series. Then at the end (question 11), adjust your deseasonalized forecast with the appropriate seasonality index the same way you did before in classical decomposition method to arrive at a seasonal forecast.

Use the deseasonalized time series for the last 4 days for questions 7 to 10 below.

7) What is a deseasonalized forecast for day 57 using the 3-day weighted moving average with weights 60%, 30% and 10%?

a. 378.31 b. 391.56 c. 384.96 d. 384.77

8) What is a deseasonalized forecast for day 60 using the 3-day weighted moving average with

weights 60%, 30% and 10%? a. 391.56 b. 378.31 c. 392.86 d. 384.96

9) What is a deseasonalized forecast for day 57 using the exponential smoothing method with

smoothing constant 0.7 (Hint: Use deseasonalized data on day 53 as your forecast for day 54)? a. 384.77 b. 384.19 c. 386.72 d. 393.28 e. 380.36

10) What is a deseasonalized forecast for day 60 using the exponential smoothing method with

smoothing constant 0.7? a. 384.77 b. 384.19 c. 386.72 d. 393.28 e. 380.36

11) Using the deseasonalized forecast of question 8 and the method of incorporating for seasonality

that the Chief Marketing Officer suggested, what is a seasonal forecast for day 60? a. 391.56 b. 384.19 c. 354.67 d. 320.51

Mini-project 3: Forecasting

pur-new-sol

Related Questions