Insight Manufacturing

Insight Manufacturing Sandra Emsai is the production manager for Insight Manufacturing, Inc. (IMI) which has only been in business for two years. IMI makes three products for the medical instrument market. The product numbers for the three products are A-307, B-299, and C-461. Until now, IMI has been relying solely on the predictions from the sales team to forecast demand for their products, and they have seen varying levels of performance from their supply chain, including some products where production and inventory levels have been fairly accurate, as well as products which seem to have large variations in the actual demand versus what was forecast, which has caused stock- outs and backorders when they swing one way, and excess inventory carrying costs when they swing the other way. Finally, Sandra has one year of reliable demand information available from her ERP system, and she would like to do a better job next year at predicting demand for these three products. Here is the demand data from the previous twelve months for the three products:

ACTUAL DEMAND

A-307 B-299 C-461

1 January 6,532.10 4,691 1,683

2 February 7,104.50 6,257 1,692

3 March 7,096.70 6,901 1,586

4 April 9,995.10 5,905 1,569

5 May 12,505.70 3,734 1,960 6 June 9,521.40 6,748 1,840 7 July 5,974.30 8,267 1,823 8 August 2,322.80 4,725 1,720

9 September 6,674.10 7,297 1,681

10 October 6,171.50 5,331 1,535

11 November 5,992.20 7,024 1,538

12 December 5,718.60 2,134 1,737

Answer the following questions from this case:

1. Analyze the historical data, and from that analysis, recommend an appropriate forecasting method for each product. For methods that require a selection of a number of periods, or

constants such as weightings or α, you can randomly select typical values for your method.

2. Using your selected forecasting methods, generate forecasts for the historical data and then use some standard measures of forecast accuracy to evaluate your methods. You can

experiment with different numbers of periods and/or weights and values of α to see if you can improve the accuracy.

3. Discuss why certain sets of data are better forecast with different methods.