Social media has become a strategic channel for regional leaders to communicate public programs and build interaction with citizens. This study aims to predict the engagement level of regional leaders' social media posts by analyzing content format, post theme, Instagram impressions, Facebook impressions, and reached accounts. The research data were obtained from the social media insight dataset of regional leaders in Blora Regency in 2026, consisting of 150 posts from 14 February to 13 April 2026. A quantitative approach was applied using multiple linear regression, while engagement was calculated from likes, comments, shares, and reposts. The categorical variables were transformed using one-hot encoding, and the data were split into 80% training data and 20% testing data. The model evaluation produced Mean Absolute Error (MAE) 163.32, Mean Squared Error (MSE) 43712.84, Root Mean Squared Error (RMSE) 209.08, and R2 Score 0.804. These results indicate that the model explains approximately 80.4% of engagement variation in the test data, although the absolute error remains relatively high because social media responses fluctuate across posts.
Copyrights © 2026