Linear Regression for Traffic Data

Hi there, I have a two full years of traffic data by month and I am trying to predict traffic for the next two years based on historical growth rates.

Previously I have used year over year and month over month growth rates to predict results and it has faired OK. Since I have two full years of data, I am also accounting for any seasonality. That said, my numbers are not very accurate and there is quite a variation on what is actually happening.

I am looking for advice/direction on whether linear regression is the right way to model this data? Should I pick a 12 month block and then create a regression and then apply a growth factor from the previous year to estimate a future month's performance?
I'd normally recommend Poisson regression given the nature of your data, but if there is seasonality effects (i.e, over dispersion), then Negative Binomial Regression might be right up your alley.
Time-series analysis is a whole sub-discipline, mostly develped by economitricians. You can read a little bit about it at After that I suggest you get a book.

Suffice to say, simple regression analysis for independent data points typically doesn't work for time series analysis, because the sequence exhibits correlations. There are a bunch of different kinds of regression specific to time series analysis, which typically go by bizarre-sounding acronyms like AR1, ARIMA, and GARCH.
ichbin & jamesmartinn -- thank you very much for your comments... you have definitely put me on the right trajectory for solving this problem and many others on my end...

first time forum user for analytics over here... incredibly impressed by this resource!:D:tup: