Algorithm behind Column Dependencies feature

Follow

Rahul Dhond

• October 16, 2015 09:55

Can anyone throw some light on the algorithm behind Column Dependencies feature in Smart Analytics? I know how the feature works in Datameer but I dont know how the dependency numbers are actually calculated and hence the curiosity.
thanks & regards,
Rahul

1

• 

  Support

  • October 16, 2015 22:16

  Hello Rahul,
  The column dependencies feature uses Mutual Information between each pair of columns: https://en.wikipedia.org/wiki/Mutual_information since it is not limited to real-valued random variables like the correlation coefficient.

  Hope this helps.

  Best,
  Jana

  0

  Comment actions Permalink
• 

  Rahul Dhond

  • October 17, 2015 04:44

  Thanks, Jana.

  0

  Comment actions Permalink

Please sign in to leave a comment.