Analysis and Prediction of Instagram Users Popularity using Regression Techniques based on Metadata, Media and Hashtags Analysis

Abstract

In recent years, social media is growing at an unprecedented rate, and more people have become influencers. Understanding popularity helps ordinary users to boost popularity, and business users to choose better influencers. There were studies to predict the popularity of posted images on social media, but there was none on the user's popularity as a whole. Furthermore, existing studies have not taken hashtag analysis into consideration, one of the most useful social media feature. This research aims to create a model to predict a user's popularity, which is defined by a combination of engagement rate and followers growth. There were six machine learning regression models tested. The proposed model successfully predicted the users’ popularity, with R2 up to 0.852, using Random Forest with 10-fold cross-validation. The additional statistical analysis and features analysis results revealed factors that can boost popularity, such as actively posting and following users, completing user's metadata, and using 11 hashtags. In contrast, it was also found that having a large number of posts and following in the past will not help in growing popularity, as well as the use of popular hashtags.

Hashtags Influence Maximization: Choosing the most Influential Hashtags on Instagram

“…In this study, two popularity metrics were used, namely Unique Engagement Rate (UER) and average Engagement Rate (ER). As commonly used in other studies, ER is defined as likes + comments divided by followers [19]. In this research, ER can be calculated for each post or each hashtag.…”

“…Users with low followers are known to produce high ER [19], which can produce unfairness for average ER calculation. In our dataset, the average ER of posts from users with 0-99 followers = 31.01%; 100-199 followers = 16.24% 200-299 followers = 13.65%.…”

“…To combine the closeness and the hashtag rank, a set of related hashtags are first selected based on closeness with the main hashtag, then be ordered based on hashtag rank, and the top N_related hashtags are the output. Note that we chose N_related=11 and N_related=30, since currently, Instagram allows a maximum of 30 hashtags in a post, while using 11 hashtags is the best for engagement rate [19].…”