Forecasting Boat Sales

You work for Lawn Boat Company, a regional manufacturer of power boats (for recreation). It is a very seasonal business, with 2/3s of sales coming in the Spring and Summer. Having a good sales forecast is essential, given the volume of sales in a single quarter (almost 40% in Spring) and lead time necessary to produce the boats. Rather than simply trying to predict sales for your company, you develop a forecast for the entire industry, and then estimate your sales as 5% of total industry sales. You have collected the following data for total sales for the power boat industry for the period 2006 to 2016, shown first for the entire year (in billions) and then broken down by quarter (in millions):

Obviously, this data includes the 2007 to 2009 recession, and the boating industry was even to slower recover than some others. This is part of the data you will use to create an annual forecast.

Besides time series forecasts, you must also try to find a good causal forecast. To that end, you have collected data (2006 to 2016) on return on the Dow Jones Index (DJI, expressed as a percentage), the Consumer Price Index (a weighted average of many prices used as a measure of inflation), the Gross Domestic Product (measured in trillions of dollars), the Prime Rate (published by the US Treasury) and the Unemployment Rate (percentage of available workers who are unemployed). This data is shown in the table on the following page.

Begin by graphing the Times Series data to look for any trends that might exist. You must explore Nave, Moving Average, Weighted Moving Average, and Exponential Smoothing forecasts. You must also explore all simple regression forecasts and, using those results and correlation analysis, explore causal forecasts via multiple regression. For each forecast, determine the error and the Mean Error (ME), Mean Absolute Deviation (MAD), Mean Percent Error (MPE), and Mean Absolute Percent Error (MAPE). Finally, using all those outputs, find the best forecast and give a forecast for 2017. Use the quarterly data in the first table to create Seasonal Indices and determine the quarterly forecasts for 2017.

GROUP INSTRUCTIONS: In addition, the groups must rework all the forecasts created in the video for just the data from 2010 to 2016 (look at the graph to figure out why those years were chosen) and using the regression outputs and error measures, choose the best forecast for the reduced (2010 to 2016) dataset and find the forecast value for 2017, again using the seasonal indices to create seasonal forecast based on the annual forecast

Finally, develop forecasts for 2018 using both the best forecast from the video and the best forecast from the reduced data set. For the causal regression, use the following estimated values from 2017: