I am trying to forecast electric power demand for natural gas in the near future. A huge driver of electric power demand for natural gas on a monthly basis is weather, specifically Cold Demand Days when you need to use your air conditioning. I have weather forecasts for Cold Demand Days by month for next year.

Also, over the past 10 years, the annual increase in electric power demand for natural gas has risen 3-5% a year as power plants use more natural gas (and less coal), and I see this trend continuing.

So…I ran a multiple regression in Excel with my dependent variable as Natural Gas Used Per Day and my independent variables as 1) Cold Demand Days and 2) a Time Variable when starts at “1” in Jan-2001 and ends with “99” in Mar-2009.

Here are the results:

Regression Statistics

Adjusted R Square 0.90

Standard Error 1.41

Observations 99

Coefficients Standard Error t Stat

Intercept 9.71 0.31 30.88

Time Variable 0.05 0.00 9.77

CDD Variable 0.03 0.00 28.76

I am not a Stats expert, so I wanted to run this by you guys to see if everything made sense. Do you think there's anything I missed? Did I handle the Time Variable correctly?

Thank you very much!