PredictG begins with an elimination round where teams choose a domain, clean messy datasets, and build optimized Machine Learning models. Top performers advance to architect complex Deep Learning solutions, tuning neural networks for high-stakes problems. Participants must deploy their final models against a hidden test set, with victory determined by prediction accuracy on unseen data.
Team Structure: Teams can consist of an individual or a maximum of 2 members. ● Format: Round 1 is an elimination round; only top teams qualify for the Advanced Round. ● Hardware: Participants must bring their own laptops with at least an i5 11th Gen processor. ● Software: Laptops must be pre-loaded with Python and standard libraries (Pandas, NumPy, Scikit-Learn, TensorFlow/PyTorch). ● Fair Play: Manual labeling, data leakage, or cheating will result in immediate disqualification. ● Winning Criteria: Rankings are based on model accuracy; ties are broken by submission time. ● Final Decision: The organizer's decision is final and binding. ● You are allowed to use exactly one AI tool (e.g., ChatGPT, Gemini) strictly for model architecture and building. ● Using AI for data cleaning or preprocessing is strictly prohibited; doing so will result in immediate disqualification without warning
