Instagram Account Classification: Influencer vs. Company Profiles

As the use of social media platforms has increased over time, there has been a corresponding rise in the number of accounts influencing masses, such as influencers and companies, active on these platforms. Particularly in marketing efforts, the use of influencers and the need for brands to identify these influencers through automated systems have increased. In this study, we developed a model using existing web scraping methods and a limited data source to distinguish between influencers and companies based on Instagram profile data. After conducting exploratory analysis on the obtained data, the focus shifted to feature extraction. To effectively utilize category-based features in the dataset, we employed the CatBoost model, achieving a %98.2 auc rate with this model. In addition to achieving success in the classification problem, SHAP values were analyzed for interpretability of the model’s results, exploring the features that distinguish brands from influencers in detail.