3
$\begingroup$

I have a problem where I want to use a classifier for it. So I defined a set of features and created a dataset. Now I want to generate some plots to understand the features. I came across the Scatterplot Matrix or the Correlation Matrix, but I'm so confused about if they are the same thing or not.

1 - Visualizing correlation matrices

2- Scatterplot Matrices

So my questions:

  • Are they the same thing?

  • Does plotting both make sense? i.e. each plot convey a different message?

$\endgroup$
1
  • 1
    $\begingroup$ You don't have to choose. Good plotting wrappers for correlation matrices, such as statmethods.net/advgraphs/correlograms.html , will happily give you the option to plot the scatters in one of the diagonals, and the coefficients in the. $\endgroup$
    – jona
    Jul 14, 2014 at 10:26

1 Answer 1

3
$\begingroup$

No, they are not the same thing.

The correlation matrix shows the strength of each correlation, but correlation is a measure of the linear relationship between two variables.

The scatterplot matrix is a matrix of scatterplots; a scatterplot lets you assess all the relationships between two variables.

The downsides of the latter are: 1) It's more work 2) It doesn't necessarily get you a single number summary (although some interesting work is being done) 3) You can't fit as big a matrix on the page.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.