Consider how well jewelry sales can be forecast on the basis of disposable personal income or unemployement rate, as measures of overall purchasing power. The explanatory variables selected are based on business logic and are ones for which data are readily available: Date - date of the sales (months); JS - Jewelry sales in millions of USD; DPI - Disposable personal income in thousands of dollars; UR - unemployment rate; We will use monthly data for jewelry sales (in millions of USD) from January 1992 through December 2004 (training set) and we want to forecast jewelry sales for each of the 12 months of 2005 (testing set). Data for the training set an be found in **JS_train.csv**, data for testing set is given in **JS_test.csv**