Historical traffic data can help network teams anticipate periods of heavier demand. A university case study compares machine-learning models for classifying daily upload and download traffic—and shows why model evaluation matters before forecasts inform operational decisions.