Hello everyone. In this video you will learn how to make a forecast and test its accuracy. Recall that we trained and built the forecasting model using 2011 and 2012 data. Now, we will use the model to make a forecast for the whole year of 2013. We shall test the accuracy of the forecast on the first three months of 2013 for which we have the actual demand data available. Before we make and test the forecast, we should first determine how to measure the forecast errors. Let the error between the observed value y_i and the predicted value y_i hat to be ei, and the average of the observed values to be y bar. We first calculate the forecast error by the mean absolute deviation, or MAD. Which equals to the average of the absolute errors. Then we calculate the forecast error in percentage by the mean absolute percentage deviation, which is MAPD, which equals to MAD over y bar. To make a forecast, we need to specify the values of the independent variables for each month of 2013. First, time in month, we need to increase the numbers sequentially from 25 to 36. For the retail price, assume the price remains the same at $799.95. For the month binary variables, for example February, or Feb, in column E, we set it to be 1 for February 2013, and 0 for the other months. Following the same procedure, we can assign the values to all month binary variables for 2013. Next, we copy the coefficients table of the model to this tab, over here. So we can use the regression equation to calculate the forecast. You can either type in the regression equation for each month of 2013 or use an Excel function of MMULT, I will explain how to generate the forecast by Excel in details in the screencast video. Given the forecast and actual sold units in the first three months of 2013, we can test the accuracy of the forecast by calculating the absolute errors, over here, using the Excel function ABS(actual value - forecast). Taking an average of these three numbers, we find the MAD = 121. And dividing the MAD by the mean demand in the first three months, we find the MAPD = 8.9%, which is very accurate. Here is the visualization of the forecasting and testing results. Where the demand data is in gray, the model in blue, and the forecast made by the model is in red. Now let's recap the course. We learned a few important lessons in this course. First, demand planning and forecasting is critical to sales and operations planning or integrated business planning and may directly affect the revenue and cost. Second, statistical forecasting is powerful. It can capture the trend, the impact of exogenous variables such as price, and seasonality. It can generate an accurate forecast to meet the needs of the operations people. Finally, statistical forecasting can identify the key drivers for demand and quantify their impacts. Thus it can meet the needs of the marketing people by letting them know how to stimulate and influence the demand precisely.