This code analyzes 43 Facebook ad sets and helps find the best ad set using the Upper Confidence Bound machine learning algorithm.

More on UCB here: http://banditalgs.com/2016/09/18/the-upper-confidence-bound-algorithm/

1. Export Ad set from Facebook ads directly for Impressions : Signups (Can use it for other data points as well).
2. Replace the data in ads-data.csv with current ads data.
3. Run: sudo python fb-ucb-ads.py

Folder contains the ads-data.csv file with Ads Impressions : Signups data.

Clearly Ad: 2 is the winner and should have higher budget allocation.

• Python 2.7
• Machine learning libraries
• Anaconda/Spyder

• Vaibhav Magon