0

I have created a dataframe with pandas, further generated a pairplot with seaborn which works fine. The problem is, that the data frame is too big, having 1700 rows and 23 columns. Therefore, the pairplots are too tiny to see anything. (besides, file size is huge, RAM is very busy).

As an example a dataframe with header and index:

    A B C ... W
AAA 1 2 3 ... 23
AAB 2 4 6 ...
.
.
.
ZZC 1700 1701 ... 1723

will lead to a pairplot of 23*23 ! Can I force to have 4 or more plot windows instead of one, when using python console? Can I utilize matplotlib for that and how?

Rockbar
  • 1,081
  • 1
  • 20
  • 31

2 Answers2

0

Is there any way to tell seaborn to divide the plot into several pieces?

not directly, as far as I know.

Maybe do subplots (5*5) with matplotlib? Thus, the data is visible?

That sounds like a feasible plan, especially since you probably want to be the one deciding which data appears on what plot. This really shouldn't take much effort to implement. Seaborn really is just something atop of matplotlib, so you can use any seaborn-specific plot type just like any other matplotlib axis.

Marcus Müller
  • 34,677
  • 4
  • 53
  • 94
  • Could you please give an example how i could combine the seaborn pairplot output with matplotlib? – Rockbar Jun 07 '16 at 15:13
0

The answer is "Yes" you can, and you should use the FacetGrid functionality. You can read about that in this tutorial. And maybe my code snippet from this answer will help you out.

Beyond that, it's hard to help with a code snippet unless you provide some sample data and be more specific on exactly you'd like your output to be.

Community
  • 1
  • 1
michael_j_ward
  • 4,369
  • 1
  • 24
  • 25