Analyzing Consumer Behavior in the Context of Online Social Media Marketing

The rapid expansion of digital marketing across social media has fundamentally transformed how consumers interact with online sellers. In Facebook Commerce (F-Commerce), understanding consumer behavior is vital for enabling targeted promotion, personalized engagement, and efficient negotiation. This study presents a machine-learning-driven framework to analyze consumer intentions using a survey dataset of 656 Facebook users. To address high dimensionality and improve predictive performance, Principal Component Analysis (PCA) is applied at multiple explained variance thresholds (75%, 80%, 85%, 90%, 95%). We evaluate three models — Artificial Neural Network (ANN), Random Forest (RF), and k-Nearest Neighbors (KNN) — on both original and PCA-transformed data. The Random Forest model yields the highest accuracy (0.98), outperforming the ANN and KNN models across most metrics. A Current Reality Tree (CRT) is constructed to identify root causes behind non-purchasing behaviors. The study contributes a novel, PCA-enhanced predictive framework for F-Commerce decision modeling and provides actionable insights for digital marketers.