Analyzing and Forecasting of Real-Time Marketing Campaign Performance and ROI in the U.S. Market

This study examines how ARIMA and Random Forest models can forecast Return on Investment (ROI) for real-time marketing (RTM) campaigns in the U.S. market. As RTM grows in importance, marketers require reliable methods to predict campaign outcomes and make informed budget decisions. Time-series models like ARIMA, configured to capture trends and seasonality, have demonstrated strong accuracy in this area, with low error rates (MAE: 0.053, MSE: 0.004, MAPE: 1.08%), which makes it well-suited for stable, trend-driven data. ARIMA's performance indicates its strength in accurately forecasting ROI for RTM campaigns, allowing marketers to optimize timing and allocate resources for a more significant impact. While Random Forest is typically effective with complex, non-linear data, it struggled with the time-dependent nature of RTM data, showing a notably higher MAPE of 37.31%. This discrepancy underscores Random Forest's limitations in predicting ROI for time-series data in RTM contexts. By combining ARIMA's strength in capturing linear trends with Random Forest's flexibility in handling complex patterns, marketers could achieve enhanced forecasting accuracy. Future research should emphasize creating hybrid models that address data inconsistencies and adapt to rapid RTM changes. These models can enhance budget allocation precision and enable more effective, data-driven strategies for impactful, real-time marketing campaign management.