The management of GSE only wants to consider the use of simple forecasting tech- niques for each product, since they carry thousands of different lines of sporting goods. As a result, attention will be limited to forecasting with moving averages. There is an obvious seasonality that is associated with this particular product, and it must be ac- counted for in the forecasts. To keep things simple, seasonal indexes should be derived from the overall proportions of monthly demand, accumulated over the entire four years' worth of data. The primary concern is the determination of the number of periods, k, that should be used in the moving-averages forecast. The objective is to find a forecasting tech- nique that is the most accurate on an overall basis. Forecast accuracy is relevant in terms of how well the actual monthly sales values compare to the seasonalized monthly forecasts. Forecasting error should be measured by the average amount that forecasts and actual demands differ. No concerns have been expressed as to whether forecasts tend to be over, or under, true monthly demands. The only interest is in the magnitude, or absolute value, of the forecast error. In order to obtain a fair comparison of how accurate the forecasts are with different k values, we will compare all different forecasts that would have been obtained when using k = 1, 2, 3...., 12 by testing them against the data set for the sequence of 36 consecutive monthly demands from January of Year 2 through December of Year 4. The demand values for the 12 months in Year 1 are needed as part of the data for getting the monthly seasonal indexes and to get the initial monthly forecasts in Year 2. However, no forecasts should be obtained for any month in Year I. The manager of GSE wants to see some specific analysis from this project. The manager wants to know which k value to use to minimize the average absolute forecast error. In addition, it is of interest to see a table of the average absolute forecast error over the range of values k = 1,2,3,..., 12. with an explanation of any particular pat- terns that are observed in this table. Once the k has been found that tends to be most ac- curate, two other things are of interest. First, we want the forecast that would be obtained for January of Year 5. Second, some evidence of how these forecasts are work- ing, with the selected k, should be provided. This can be done by developing a plot that shows the 36 monthly demand values from January of Year 2 through December of Year 4, along with the 36 associated forecasts that would have been obtained with the k that has been selected. The management of GSE only wants to consider the use of simple forecasting tech- niques for each product, since they carry thousands of different lines of sporting goods. As a result, attention will be limited to forecasting with moving averages. There is an obvious seasonality that is associated with this particular product, and it must be ac- counted for in the forecasts. To keep things simple, seasonal indexes should be derived from the overall proportions of monthly demand, accumulated over the entire four years' worth of data. The primary concern is the determination of the number of periods, k, that should be used in the moving-averages forecast. The objective is to find a forecasting tech- nique that is the most accurate on an overall basis. Forecast accuracy is relevant in terms of how well the actual monthly sales values compare to the seasonalized monthly forecasts. Forecasting error should be measured by the average amount that forecasts and actual demands differ. No concerns have been expressed as to whether forecasts tend to be over, or under, true monthly demands. The only interest is in the magnitude, or absolute value, of the forecast error. In order to obtain a fair comparison of how accurate the forecasts are with different k values, we will compare all different forecasts that would have been obtained when using k = 1, 2, 3...., 12 by testing them against the data set for the sequence of 36 consecutive monthly demands from January of Year 2 through December of Year 4. The demand values for the 12 months in Year 1 are needed as part of the data for getting the monthly seasonal indexes and to get the initial monthly forecasts in Year 2. However, no forecasts should be obtained for any month in Year I. The manager of GSE wants to see some specific analysis from this project. The manager wants to know which k value to use to minimize the average absolute forecast error. In addition, it is of interest to see a table of the average absolute forecast error over the range of values k = 1,2,3,..., 12. with an explanation of any particular pat- terns that are observed in this table. Once the k has been found that tends to be most ac- curate, two other things are of interest. First, we want the forecast that would be obtained for January of Year 5. Second, some evidence of how these forecasts are work- ing, with the selected k, should be provided. This can be done by developing a plot that shows the 36 monthly demand values from January of Year 2 through December of Year 4, along with the 36 associated forecasts that would have been obtained with the k that has been selected