Analysis and Auto Label

[andreaschandra]

Checking outlier of both hashtag and mention count. from this, we can automatically label account from threshold max or above average as a buzzer

This distribution will be combined with initial 200 accounts and 7000 accounts

[andreaschandra]

and check if 7200 accounts using this threshold is effective and accurate enough.
After apply thresholding, checking manually if there is a different label

[andreaschandra]

combination 46K Users + 7200K users

[andreaschandra]

In total, we have 41,756 Users

[andreaschandra]

remove account that has less than 100 tweets, end up we have 39,343 Users