I'm actually doing an economics paper on the topic of peak oil just now but I am struggling with statistical analysis and was wondering if anyone could help me.
Hubbert's peak oil theory basically says that world oil production will increase through time until a point where it 'peaks' at a maximum and will then enter terminal decline thereafter. Hubbert initially proposed a bell curve as the estimate of oil production, later revising this to a logistic distribution.
My problem is that I'm pretty uneducated in the field of statistics and econometrics and I'm sort of feeling my way in the dark in this area.
If I understand correctly, the Hubbert curve is the probability density function of the logistic curve, and the logistic curve itself serves as a 'running total' for oil production.
I basically want to construct a function, which will take the form of a best fit line for the data, so that I can check whether they best fit a normal, logistic, Cauchy, etc distribution. My ideas regarding this were to look at the kurtosis that the function exhibits (excess kurtosis > 1 means it is closer to logistic than normal, I think?)
Would I be able to draw 'model' bell curves, logistic PDFs etc alongside my estimated function as I know the area underneath the graphs?
At this point in time, peaking has either (a) occurred already or (b) is imminent. I want to construct this function so that it will have some predictive power about this peaking and future decline in production. I can calculate the total recoverable oil (which I presume is the area underneath the oil production graph). Will this be useful in the calculations?
Thanks ever so much if anyone can help!
Peak oil statistics watch
- Thread Starter
- 04-02-2010 21:12