NAME: Electric Bill Data
TYPE: Sample
SIZE: 120 observations, 13 variables
DESCRIPTIVE ABSTRACT:
The dollar amount for a monthly (January 1991 through December 2000)
household electric bill is presented as a time series. In addition,
potential explanatory variables are included. Twelve representative
monthly values are provided for the average temperature, for
heating degree days, and for cooling degree days (not for each
month for each year). Additional variables give the family size
each month and indicate when a new electric meter and new heating
and cooling equipment was installed. To convert the billing amount
to estimated power consumption, a tiered rate function (supplied
in the accompanying Instructor's Manual) and the costs of
associated riders (provided here) must be used. Consumption
estimates resulting from this information are supplied.
SOURCES:
Personal data records were used for the actual billing amount and
other household variables. Temperature and heating and cooling
degree days can be retrieved from NOAA sites such as
http://lwf.ncdc.noaa.gov/oa/documentlibrary/hcs/hcs.html. Heating
and cooling degree days computation methods were revised in July
2002.
DATASET LAYOUT:
Column Description Label
1 - 3 Observation number NUM
5 - 8 Year YEAR
10 - 12 Month MONTH
14 - 19 Amount of bill (in dollars), BILL
includes 5% sales tax
21 - 24 Average temperature (in degrees Fahrenheit) TEMP
26 - 29 Heating Degree Days HDD
31 - 33 Cooling Degree Days CDD
35 Number of family members at home SIZE
37 New meter? (indicator variable, 1 = yes) METER
39 New heat pump 1? (indicator variable, 1= new) PUMP1
41 New heat pump 2? (indicator variable, 1= new) PUMP2
43 - 52 Total charge (per kwh) for all riders RIDER TOTAL
54 - 58 Calculated consumption (in kwh) CONSUMPTION
The dataset contains values for January 1991 through December 2000.
The values are aligned and delimited by spaces. Missing values are
denoted by "*."
SPECIAL NOTES:
The entry for the BILL in January 1994 is missing. The entry of $0.00
for August 1999 is correct. This value prompted the power company to
replace the electric meter.
Heating Degree Days is defined as the cumulative number of degrees in
a month by which the mean temperature falls below 65 degrees. These
values are thirty-year averages for this geographic location.
Cooling Degree Days is defined as the cumulative number of degrees in
a month by which the mean temperature rises above 65 degrees. These
values are thirty-year averages for this geographic location.
Twelve values are provided for Average Temperature, Heating Degree Days,
and Cooling Degree Days. These values repeat over the course of the
time series.
PEDAGOGICAL NOTES
These data are appropriate to use in statistics classes at a variety
of levels. Using only the billing amounts, the data provide a time
series for students to examine visually for seasonal patterns and
trend. Analysis could lead to the discussion of the treatment of
missing values and outliers, but if this is beyond the scope of the
course, the instructor could certainly adjust the values for those
periods prior to giving the data to the students. This time series
is quite effective for teaching seasonal decomposition and other
forecasting techniques. When additional variables are incorporated,
it is a good application for multiple regression. The data can also
be used for spreadsheet exercises.
SUBMITTED BY:
Constance McLaren
Analytical Department
Indiana State University
Terre Haute, IN 47809
USA
c-mclaren@indstate.edu
Bruce McLaren
Organizational Department
Indiana State University
Terre Haute, IN 47809
USA
b-mclaren@indstate.edu
--