Comparison of Correlation results of multiple users

Viewed 24

How can I create correlation heatmaps of multiple users by 'Label' column. I want to compare all of them at once.

Would be helpful if I could filter values where (absolute) correlation is above 90%.

Index(['DATE', 'Creation_Date', 'Target_CPA', 'Site', 'Status', 'Label',
       'Campaign_Name', 'Campaign_Query', 'COST', 'REVENUE', 'AdWords_Clicks',
       'AdSense_Clicks', 'AdSense_IMPR', 'Adwords_IMPR', 'Conv', 'AdWords_CPC',
       'AdSense_CPC'],
      dtype='object')

Currently, I'm using this code by creating separate csv for each Label row value-

df.corr().unstack().sort_values(ascending=False).drop_duplicates()
 
Result:
Target_CPA      Target_CPA        1.000000
AdWords_Clicks  AdSense_IMPR      0.996090
Conv            AdSense_Clicks    0.980790
COST            AdWords_Clicks    0.963979
AdSense_IMPR    COST              0.959211
0 Answers
Related