Decoding Super Bowl Ad Popularity features
This project is done as part of the course DS5220 - Supervised Machine Learning
Analyzed a dataset of ads from 10 brands and predicted ad popularity using a fine-tuned random forest model and found out the key features for popular ads include animals, sex appeal, and humor.
We are seeking to find out what the driving force is behind the popularity of Super Bowl commercials. The Super Bowl provides an opportunity for brands to promote their products with unique and often over-the-top commercials, so we want to discover what the defining characteristics are for a successful Super Bowl ad.
Brands pay millions of dollars for an ad during Super Bowl commercial breaks, and a significant portion of viewers enjoy the commercials just as much as the actual sporting event. Viewers often rank their favorite commercials and brands can truly make significant profits if they can put out a hit commercial. Thus, it is valuable to see which combination of absurdity, hilarity, danger, sex appeal, etc. enthralls the most viewers.
Our dataset consists of 247 advertisements from the 10 brands that aired the most commercials this century. Each row represents a different advertisement. Columns describe different variables relating to each advertisement. Variables range from binary descriptors (funny, patriotic, etc.) to video performance and descriptors (view count, like count, comment count, video title, channel title).
