This capstone project developed an integrated social media analytics dashboard for Parker Marketing & Management (PMM), a marketing agency serving clients in the medspa and wellness sectors. The objective was to transform fragmented social media and digital engagement data into a unified, predictive decision-support dashboard. By leveraging platform APIs—including Facebook, Instagram, LinkedIn, Google Analytics, and Mailchimp—the project built a Python-based ETL pipeline to extract, standardize, and aggregate daily metrics into a dimensional star schema. 

To address volatility and zero-inflation in social media data, the modeling strategy adopted a two-part Random Forest approach: first, classifying whether a platform-client pair would generate engagement in a given week, and then estimating the engagement magnitude only when activity was likely. Feature engineering incorporated calendar cycles, rolling averages, lagged metrics, and spike detection to improve prediction quality. Despite strong in-sample model performance, limited historical data reduced out-of-sample reliability, emphasizing the importance of sustained data collection and quarterly retraining – for future implementation. 

The project culminated in a responsive Dash-based dashboard that visualizes historical trends, forecasts, cross-platform correlations, and actionable posting recommendations. Confidence metrics such as R² are integrated into every forecast visualization to maintain transparency. The system bridges technical analysis and practical usability, helping account managers plan campaigns, evaluate performance, and justify marketing decisions with greater confidence. As more data accrues, the platform is positioned to evolve into a fully automated, client-specific analytics and forecasting tool that supports long-term growth and strategic optimization for PMM’s digital campaigns.