Benchmarks¶

These reports compare deep-river models against strong streaming baselines on standard datasets. Use them to understand trade-offs between predictive quality, model complexity, and update behavior in online scenarios.

Each benchmark chapter includes dataset context, model configurations, and visual summaries to support reproducible comparisons.

How to read these reports¶

• Treat the tables as task-specific comparisons, not universal model rankings.
• Compare predictive metrics together with memory and update time; deep models often trade speed for flexibility.
• Pay attention to stream order and dataset context, because online results depend on when each item arrives.

Reports¶

• Binary Classification
• Multiclass Classification
• Regression