I.T. Times
Volume 4. No 4 Information Technology News of the University of California, Davis December 1995


Predicting the Car's Future

Researchers Develop Model to Predict Consumer Behavior

by Anne Jackson, Information Technology Publications


Will Californians embrace the electric car? Who is most apt to buy alternative fuel and electric vehicles? And how many electric car owners are likely to reside in each region of the state?

Public utilities would like to know the answer to those questions, since they estimate that an electric car sitting in the garage charging away each night will require as much electrical power to run as an entire house.

To plan future power plant facilities, utility companies have to be able to pinpoint where the need will occur, so they are anxious to find a way to factor in how many people will be buying electric cars. The difficulty is coming up with a way to predict who will buy a product that exists, for the most part, only in prototype.

Enter David Bunch and a team of interdisciplinary researchers from the Institute for Transportation Studies. Bunch, of the UC Davis Graduate School of Management, together with Tom Golob and David Brownstone of UC Irvine, has put together a microsimulation model to project California vehicle purchases over the next few years. The project has received major funding from the California Energy Commission, Southern California Edison, and Pacific Gas & Electric, and has generated interest from national and regional public policy planners, as well as auto manufacturers.

Driving the project is a mandate from the California Air Resources Board (CARB) that zero emission vehicles amount to a specified and gradually increasing percentage of vehicles purchased in California. The required percentages range from 2% in 1998 to 10% by 2003.

"We're using a quantitative approach to try to mimic what the future vehicle market will do by simulating vehicle purchase behavior at the individual household level," says Bunch.

"What distinguishes the project," he says, "is that it is a state-of-the-art model based on sequences of vehicle transactions, whereas most existing models provide a snapshot of a cross-section of households and deal with what a consumer will grab from what is on the shelf." Bunch's model divides time into 6-month increments to simulate events occurring at 6-month slices in time.

The model is built on a sequence of surveys beginning in 1990, which gathered detailed demographic information from 4,747 selected individual households. The survey examined past vehicle purchases and took subjects through a series of preference tasks, asking them, for instance, which of several hypothetical vehicles they might purchase. That initial survey was updated a year later with responses from 2,900 households and will be refined further in 1996.

The group used a Matlab-based microsimulation system to construct the model. The portability of the Matlab system appealed to the researchers because it makes it possible to develop the model on one platform and move it to another.

"We have these models running on workstations," says Bunch. "Now the challenge is to analyze the data after we produce it. The data can be aggregated up to any level of geographic detail."

"At this point," says Bunch, "we have a sample of households and we're applying weights and measures to get a picture that can be summarized up to the level of the entire state of California."

The project's public utilities clients can run the Matlab forecasting model, take the output database from the model, and use it to output to a Geographic Information System (GIS), where they can display it on a map for planning at the level of a utility planning district.

Now Bunch and his colleagues are at work producing similar models for planners who report energy forecasts to the state legislature for different regions of California.

Travel the Web for More Information

Further information on transportation studies may be found on the World Wide Web. Here are a few locations that may be of interest to you.

Bureau of Transportation Statistics: The Smart Library
http://www.bts.gov/smart/smartlinks.html

Integrated Transportation Management Center
http://herman.tamu.edu/itmc-mission.html

Transportation: Engineering and Technology
http://www.einet.net/galaxy/Engineering-and-Technology/Transportation.html

California Smart Traveler
http://www.smart-traveler.com/


ietpubs@ucdavis.edu