We noticed that in both the person level section, and the household level section, there are 34,346 households that have oil heat.  That tells us that every house that is heated with oil has occupants.  We would expect to find some percentage of vacant homes to have oil heat.

If we had a lesser tool than we do, we might have to accept that all oil heated homes were occupied on census day. Or more likely, with a lesser tool, we would not have even noticed this problem.  We know of no other tool that does analyses at both the person and household levels, simultaneously.

But with our tools, it is a simple matter to look into the situation.  We wonder what kind of fuel the vacant houses are using.  Note that in the earlier analyses, we had recoded the heating fuels to show only electric, oil, and "other".  Below is an analysis of puma 3200, with a full breakout on heating fuel, and also a dimension of "number of occupants".  This analysis is just as we would give it to the customer, comma delimited to be fed into a data management tool.

Persons,Fuelheat,Houswgt
'0-0,N/A (GQ/vacant), 60199
'1+,Gas:  Underground pipes, 6443
'1+,Gas:  Bottled  tank  or LP, 3699
'1+,Electricity, 23438
'1+,Fuel oil  kerosene  etc., 34346
'1+,Coal or coke, 8066
'1+,Wood, 5353
'1+,Solar energy, 36
'1+,Other fuel, 119
'1+,No fuel used, 54
Persons,Fuelheat,PHouseholds
'1+,Gas:  Underground pipes, 6443
'1+,Gas:  Bottled  tank  or LP, 3699
'1+,Electricity, 23438
'1+,Fuel oil  kerosene  etc., 34346
'1+,Coal or coke, 8066
'1+,Wood, 5353
'1+,Solar energy, 36
'1+,Other fuel, 119
'1+,No fuel used, 54

We can see by the above numbers that the Census Bureau did not identify the heating fuel for vacant houses.  This is a problem for our oil dealer, because there are a lot of vacant houses, and some percentage of the houses that are vacant do use oil, but they probably use less oil than the occupied homes do. 

If the customer wanted to, we could at this point perform more simulations to account for the unidentified homes.  For example, we could determine what percentage of homes in the area use oil heat (based on the homes which we know to have an identified heat source), and we could bump up the existing oil heated homes by the appropriate factor.  We could account for the fact that these homes use less oil (since they are vacant) and bump down the usage of oil by an appropriate amount.  But since this is an example, and we've already demonstrated the simulation capabilities, we won't go into that here.