I've a trainingset which has 400 features and most of them have null value.

I tried to draw the heatmap of nullity correlation matrix by means of Python and missingno, but the heatmap is unreadable due to high number of features.

How can I print the nullity correlation matrix, instead of draw it?

  • $\begingroup$ Hi, and welcome! What framework are you currently using? If it can draw the heatmap, surely you can retrieve the values somehow. $\endgroup$ – Romain Reboulleau Oct 19 '19 at 6:09

Using pandas, the nullity correlation matrix seems to be obtained by df.isnull().corr() (this is how it is done is missingno), and this makes sense.

missingno package also states, in the heatmap function documentation, that for large datasets the dendogram view is better. However, it does not tell me if "large" means many features or entries.

|improve this answer|||||

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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