Predicting User Tipping

Project overview:

• Preprocessed 100 GB of tweets using pandas and extracted new insights into norm dynamics leading to paper (Alharbi et al., 2024).
• Developed a novel graph-based model (TGNN) using NumPy and PyTorch to predict when users will adopt certain social norms, successfully identifying susceptible users with an AUC of 0.95 leading to second paper (Kashuv et al., 2024).
• Solely responsible for all aspects of the project, including problem formulation, methodology design, data pre-processing, implementation, model training, and performance evaluation.