3
$\begingroup$

I'm trying to visualize data for exploratory data analysis motivated by visualizing multiple scatterplots of features simultaneously, similar to this question. But I quickly run into problems when using a large number of features (~50) and rows (~50K). While I like using seaborn pairplots the generation of a large number of plot panes can get computational intractable for a large number of features and observations. Subsetting a very large table to a smaller number of features or observations does not seem complete.

My question is: What's an efficient way to plot many features for EDA in python? If there is not an efficient way then is there a defensible way to reduce the number of features or observations

Improve this question
$\endgroup$

1 Answer 1

Reset to default
-1
$\begingroup$

I would suggest t-SNE (just google it). It helps you to have a general overview on what is gling on with your high-dimensional data. Please note that parameters of t-SNE are pretty sensitive so you need to put some effort to find a good embedding.

Improve this answer
$\endgroup$

Your Answer

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

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.