Influencer Identification System Design Using Machine Learning Techniques

Abstract

Being one of the most effective spreading methods, word-of-mouth networking moved to its digital version - social media platforms. Social media represents a digital platform which allows users to interact with each other in different ways. Audio, video, image or text content are examples of such interactions. Widespread use of social media platforms brings new ways to marketing solutions in order to reach potential audience, especially thorough those users who actively interact with and have an influence on their followers. Brands which utilizes abilities of such users of social media are involved in field of influencer marketing. One of the challenges in this rapidly growing field is finding person (influencer) which will match requirements of the brand. Rising number of influencers and social media platforms leads to increasing data and difficulties in finding appropriate influencer. The purpose of this study is to provide design of an influencer identification system utilizing machine learning algorithms. Although the number of Influencers rising, they are still in minority comparing to all users of social media. Thus, focusing on solving problem of imbalanced classification, performances of Logistic Regression, Linear Discriminant Analysis, K-Nearest Neighbors, Support Vector Machines and different tree based methods were compared and Random Forest method is selected to be used in the system.