Logistic regression is based on logarithms. Ordinary Least Squares regression and analysis of variance uses the actual values as the dependent and independent variables in an equation. Logistic regression does not.

What is a log, anyway?

Let’s start with the very basics. First we learned to add:

5+5+5+5 = 20

After about eight years of age, we realized that was pretty inefficient so we started multiplying

5 x 4 = 20

We got a few years older, thought, why stop there, and got into exponents

5 x 5 x 5 x 5 became 5 to the fourth power = 625

Then we get into logarithms, where the log to the base 5 of 625 = 4


Think about this. Really think about it. Go to the wikipedia page that has a good explanation of logarithms and read that.

Calculate the logs of several numbers to different bases, just for the heck of it. I have noticed that, so often, students skip over topics like logarithms thinking, “I don’t need to know that.”

This is wrong on a whole lot of fronts, just one of those reasons being it is a really bad habit to get into. I don’t know how many reports I read in the newspaper of people losing their homes that included the statement,

“Mr & Mrs John Q Public said that they did not understand the mortgage papers, that they just trusted the real estate agent, the banks or the ad they watched on TV at 2 a.m.”

So, whose fault is that. Understand what you are doing! Start with logarithms. It’s as good a place as anywhere else.

Similar Posts

One Comment

  1. I named my blog logbase2 as a kind of play on words, only to find that a lot of people wanted to know what a base 2 logarithm is all about.

    I have written a couple of posts on the subject.

    I think they have a pretty good half-life.

Leave a Reply

Your email address will not be published. Required fields are marked *