Study Guide for Blockhesd Concrete

The spreadsheet Foundation Blocks uses seasonal decompoditiom time series to forecast Blockhead Concrete Company sales of concrete blocks designed for building foundations in the same way Rocky Gold used seasonal decompoditiom time series to forecast jewelry sales.   The purpose of the forecast is to help Blockhead decide how much inventory to produce and stock.  The details are in the link Review Seasonal Time Series.

The sppreadsheet Landscape Blocks is mainly concerned with forecasting sales of concrete blocks designed for use in landscaping.  While builders have to buy foundation blocks very early in the building process, they buy landscape blocks only when they are nearly done.

Tab "Seasonal" of spreadsheet Landscape Blocks uses the by-now-familiar seasonal decompoditiom time series method to forecast sales of landscape blocks.  Note that the MSE is 330,573.

Tab "Leading" of spreadsheet Landscape Blocks uses Housing Starts as a leading indicator to forecast sales of landscape blocks.  See page 1 of the text for a discussion of leadind indicators.  An indicator (or "Causal") model is a regression model where X is the indicator variable used to forecast the following month'svalue of the variable Y you're interested in.  In a "leading indicator model, past or present values of X are used to forecast future values of Y.  In the example in the tab "Leading," the number of housing starts this month is used to forecast the number of landscape blocks needed next month as the houses begun this month near completion.  The MSE is 262,121, indicating better performance thab the time series model; this is not surprising because the jeading indicator model incorporates more information.

Tab "Simultaneoua" of spreadsheet Landscape Blocks uses Housing Starts as a simultaneous indicator to forecast sales of foundation blocks in the samr month as the housing starts.  The MSE is 233,208 which is worse than the MSE of 136,730 for seasonal decompoditiom time series from the spreadsheet Foundation Blocks, but even if the Msefor the simultaneous indicator were much better it would be useless for Blockhead's inventory management becauae the Census Bureau doesn't publish the Housing Starts data until after the end of the month.

We have already looked at a more useful example of  a simultaneous indicator model, model 2 in  Pie Processing Cost; see pages 40-42 of the textbook.   Since the number of pies he's going to bake in the coming week is Mr. Simon"s own decision, he knows it ahead of time and can use it to forecast his processing cost for that same week.

For exams, you will have to read Excel output like lines 83-100 tab "Simultaneous" of  spreadsheet Landscape Blocks and pick out the slope and intercept of the indicator model.  In the example, the slope is 0.098 thousand additional foundation blocks per  housing start and the intercept is 228.781 thousand landscape blocks.  You will also have to "plug in" a specified value of the X variable ti firecast the corresponding value of the Y variable; for example, if there are  20,000 housing starts in a  oarticular month we would expect  228.781 +  0.098*20,000 = 2,192.337 thousand landscape blocks to be demanded that same month.