travel-code-sleep/Reinforcement-Learning-Using-UCB-and-Thompson-Sampling

Find the best advertisement to show on multiple websites while still exploring different Ad options. Save time and money that would have to multiple AB testing. Start getting rewards early.

Reinforcement Learning Using UCB and Thompson Sampling

Created a Reinforcement Learning Model to choose the best advertisement to display on several websites during an Ad-campaign. Used Upper Confidence Bound algorithm to create a baseline and Thompson Sampling to improve the online learning model that came up with the best ad from a group of 10 similar advertisements in far less time, which is required for multiple AB tests, and maximized click through rate.

Attached Data File is a preprocessed simplified file that contains rewards as columns for each ad. Each index/row is a round in which those 10 Ads were displayed. Reward columns have 1 if the ad is clicked during that round else 0.