Determining Predictability of Sales Orders from Discount Requests

This is a case study which analyzed whether a discount request will ultimately turn into a sales order for Greenheck Fan Company.  To do this analysis, the discount request database as well as the sales order database were analyzed and compared to determine predictability of an order.  By being able to predict a sales order, this will create more efficiencies for employees as well as aid in predicting demand for products will also help with staffing and investment opportunities.  The data was cleaned and then processed through both linear and logistic regression models to see if an order can be predicted from the data.  The variables that were utilized were mostly text variables that applied to different pricing categories as well as manufacturing competitor names.  By analyzing which pricing categories were utilized, the competitors listed, basis of design, and office numbers there were somewhat accurate predictions that were made by the models.  Real data was pulled and compared to the created model and then tested for accuracy.  The logistic regression model performed the best due to its categorical nature and was able to predict an order at a fairly high rate.  The models will remain as a supplement to other forecasting methods that are already implemented at Greenheck but will continue to be monitored for accuracy.  Also, these models will be reworked to add in other variables that may make them even more effective.