NEXT-GENERATION ENGAGEMENT FORECASTING FOR INSTAGRAM USING EXPLAINABLE MACHINE LEARNING MODELS

Abstract

The increasing popularity of social media platforms, especially Instagram, has reshaped online communication and digital marketing, where user engagement now serves as a key indicator of content performance. Instagram has evolved from a simple photo-sharing application into a highly interactive platform where businesses, influencers, and organizations depend on engagement metrics such as likes, comments, shares, and reach to assess content performance. As the volume of content and user interactions continues to grow, analysing and predicting engagement rates has become increasingly challenging due to the complex relationships among user behaviour, content characteristics, and platform dynamics. The primary problem addressed in this research is the difficulty of accurately predicting Instagram engagement rates using traditional analytical approaches. Conventional methods rely on platform-generated statistics, manual analysis, and static reports, which provide descriptive insights but lack predictive capabilities. These approaches are often limited by poor scalability, inability to identify hidden patterns, delayed decision-making, and dependence on subjective interpretation. To overcome these limitations, the proposed system implements a machine learning-based framework for predictive modelling of Instagram engagement rates. The framework includes data preprocessing, exploratory data analysis, feature scaling, model training, and performance evaluation. Several regression algorithms, including Linear Regression (LR), K-Nearest Neighbours Regressor (KNNR), Random Forest Regressor (RFR), and Decision Tree Regressor (DTR), are developed and compared using evaluation metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Square Error (RMSE), and R² Score. Experimental results indicate that tree-based models outperform other techniques, with the Decision Tree Regressor achieving the highest prediction accuracy. The proposed system provides an effective, scalable, and data-driven solution for engagement prediction, supporting improved content strategies and informed decision-making in social media analytics.