1. Gulley's s Ice Cream Parlor wants to forecast ice cream sales. The company believes that sales are dependent on the weather temperature (i.e.: the hotter the weather the more ice cream sales, etc.). The data required for this forecast is to determine an average daytime temperature say for over a 15 week period and also for each of those weeks determine the amount of ice cream sales. The forecasting model that should be used is the Linear Regression model. In this model the variable being forecast (ice cream sales) called the dependent variable is related to some other variable (temperature) called the independent variable. Within this model, the correlation coefficient needs to be computed. This coefficient measures the strength of the relationship between the variables. In this model, the coefficient should be +1 which means that there is a positive linear relationship. When the weather gets hotter or increases so does the sales of ice cream increases. All models are not 100% accurate; therefore, it is necessary to measure forecast accuracy by using either MAD or MSE methods or preferably both. Assessing this forecast model alone is difficult. The lower the value of MAD relative to the magnitude of the data, the more accurate the forecast. Therefore, to achieve a better forecast another forecasting model would have to be used to evaluate and compare the MAD and MSE values and to assess which model is more accurate.

2. Three Tier Candles & Body Treats should forecast the demand during a specific period of time to determine the weighted average to forecast needed inventory during another period. Using this model will allow the owner to determine the amount of product needed to fill a specific forecasted demand. Understanding this data could allow the owner to better utilize JIT principals regarding product inventory and the amounts needed to have on hand as a consistent stock. To begin this forecast, the owner would "use the naïve method to generate an initial forecast" since an initial forecast is not available (Reid & Sanders, 2010, p. 280.). Once data is complete and graphed, to determine the most stable variation, the forecast with the least variation would be used.