# How can I make clusters with highly right skewed data?

My histogram of expected income is as below.

As you can see, income is highly right skewed. I want to divide individuals with regard to income - for example into Low, Middle, and High, and then I want to compare the other properties of individuals in each of the three groups.

Any idea about dividing into groups with highly right skewed data?

• Take the logarithm of x and then plot that. Jul 30 '14 at 5:56

Use a non-linear function. Functions such as $\log x$, $e^x$ and $\sqrt{x}$ change skewedness of your data.
For income data, $\log(x)$ or $\log(x+1)$ might work.