Predicting the price of gas

May 01, 2003, vol. 27, no. 1
By Carol Thorbes



Document Tools

Print This Article

E-mail This Page

Font Size
S      M      L      XL

Related Stories

Like most consumers when they fill up their car, Crystal Linkletter is more worried about the price of gas than how much of the liquid gold is in a barrel.

However, the Simon Fraser University student's ability to predict the amount of fuel (such as gasoline or natural gas) a producer has on hand, despite reporting delays, has impressed two national bodies in the United States.

For her master's in science thesis, Linkletter determined the best method by which Texan natural gas producers could calculate their product on hand, without final production figures.

The Energy Information Administration (EIA) and the American Statistical Association's (ASA) advisory committee may adopt Linkletter's methodology. They were impressed with her thesis presentation in Washington D.C. in April.

The EIA conducts and analyses surveys and produces models on energy use, production, availability and cost in the United States.

“Many researchers use the EIA's published production figures,” says Linkletter's thesis supervisor, SFU statistics and actuarial science professor Randy Sitter, who accompanied her to Washington. “The EIA's primary client would be Congress, which needs the figures for policymaking. Economists would also be interested in them.”

Reporting delays can throw off estimations of the amount of energy products on hand. The EIA found its early predictions were consistently under final figures in Texas.

The problem prompted the organization, in conjunction with the ASA, to award Linkletter a $12,000 U.S. fellowship a year ago, after reviewing her research interest.

Linkletter's assignment was to help the EIA improve its monthly predictions of natural gas production in Texas in the face of three-month to year-long delays in getting confirmed production figures.

“I reviewed predicted and confirmed figures over a four-year period and found they varied by the same amount roughly,” explains Linkletter. “Based on that difference, I developed an adjustment factor which when applied to more recent predictions made them better match final figures.”

Linkletter is fascinated that the analytical technique that works best for the EIA is also used to estimate AIDS cases and product failure rates while under warranty in the face of reporting delays. “This type of interdisciplinary work is a hallmark of our program in applied statistics at SFU,” says Carl Schwarz, department chair.

Search SFU News Online