An analysis and prediction model of outsiders percentage as a new popularity metric on Instagram

Abstract

In this research, a new Instagram popularity metric was defined, i.e. outsiders percentage (OP) of a post. Outsiders are non-followers who liked a user’s post. It was found that OP is the most effective metric if compared to engagement rate and followers growth. Regression models were tested for predicting OP, using features from user data, post data, hashtag, engagement, and image sentiment. The prediction accuracy (R2), reached up to 71.9% using Random Forest. This research also analyzed the trend of each feature against the OP. It was found that hashtag usage is the most important factor in raising OP.

Keywords

• Social media
• Instagram
• Popularity
• Regression
• Machine learning