Heating US
[/home/heating/us]
- state or region in the US
- type of home: attached, detached, apartment
- number of bedrooms
- heating fuel
US RECS
The US Residential Energy Consumption Survey RECS publishes data based on questionnaires returned by residents in the USA. The most recently published survey was conducted in 2001 (though 2005 data has recently become available) and contains data on 4,822 homes. In what follows, each home in the survey is referred to as an entry.
The number of entries limits the options we can offer in CarbonKit whilst maintaining reliable statistics. For example, if state, home type, bedrooms and fuel each have 5 options, then there are 5x5x5x5=625 possible selections, giving an average of 4,822/625 = 7.7 entries per selection.
The most recent RECS survey was in 2005, but this data is still being processed and has not yet been published.
Survey data
The survey data is split up into several data files. The ones used here are:
- household characteristics
- housing unit characteristics
- space heating
- energy consumption
State or region
Survey data is available for the four largest states (LRGSTATE):
- New York State
- California
- Texas
- Florida
Home type, size and age
The home type used in CarbonKit closely follows the survey's house types (TYPEHUQ), except that the survey's two types of apartment (based on size of apartment block) are combined in CarbonKit.
The number of bedrooms (BEDROOMS) is collapsed into 4 options in CarbonKit:
- 0 or 1
- 2
- 3
- 4 or more
Fuel
The FUELHEAT column in the "space heating" data file specifies the fuel used for space heating of the home. There are only sufficient entries to obtain reliable statistics for natural gas, electricity and fuel oil. Availability of these options is state dependent, for example, fuel oil only appears for apartments in New York state.
Note: The equivalent of FUELHEAT for water heating is not used in the CarbonKit processing just now. It is assumed the fuel for water heating is the same as for space heating - probably OK in most cases, but not all. The survey contains details on the heating system used, but there was insufficient data to obtain reliable statistics on this given the subdivisions of state, size and fuel.
Data processing for CarbonKit
Java classes are used to load, process and output the results into CSV files. For each possible selection of home options - e.g. texas, detached, 3 bedrooms, natural gas - the survey data was searched for matching entries in the survey. Selections with less than 5 matching entries are deemed to have insufficient data and so will not be used in CarbonKit.
For options with sufficient entries, a mean is calculated over all the matching entries for the amount of fuel used. The mean is weighted using the survey's weight factors that indicate how common each entry is in US housing stock. Separate means are calculated for the amount of fuel used for space heating, water heating and air conditioning.
If a value is zero, then it is not used to calculate the mean and is classed as a "bad" value.
Note: This seems to occur in cases where a different type of fuel is used for space and water heating, or where a house has no air conditioning. This will be accounted for properly in future versions of the processing code.