I'm teaching a Management Information Systems course for business students this semester, and we're starting a discussion tomorrow about technology trends including the famous Moore's Law -- the observation by an Intel engineer that the number of transistors on a microcontroller chip was effectively doubling every 18 months. I want my students to understand what that kind of a growth function means, so I put together a spreadsheet tutorial for them. In it, I walk them through a simple example of linear regression using the LINEST function with some linear data.
I then have them use some data from Intel to plot out the nearly exponential trend in their transistor density growth. I show them how to use the LN function to perform a natural log transformation on the exponential data to make it more linear so that we can use LINEST to calculate a linear model. I then compare the results of that linear model with the output of the LOGEST function. The emphasis here is to show that the slope of the linear model is the same as the natural log of the growth factor b in the equation y = a * b^x. I then discuss how we can approximate the doubling rate by dividing 70 by the growth rate. In this example, the growth rate is about 34% per year, which results in a doubling every 2.05 years.
UPDATE: The link I had previously somehow was missing a page of instructions. The revised materials are available at http://public.iwork.com/document/?a=p50683115&d=Transformations_and_TrendLines_Tutorial.pages
