Trendy cen biletow lotniczych - dane anonimizowane, usrednione per kategoria
Problem: Flight prices change dynamically, and popular "tips" (e.g. "buy on Tuesday") are often a myth.
Solution: I built a system that tracks thousands of price observations and uses AI models (XGBoost) to identify real pricing cycles for specific routes.
Conclusion: The tool not only shows history but indicates the optimal time window to purchase, avoiding the "rush tax".
Dataset statistics — number of routes, observations and cohorts, plus price changes in 1/3/7-day windows.
AI-generated price forecasts for the next 3-90 days before departure.
Prognoza trenowana codziennie o 02:00. Horyzont: 3-90 dni przed wylotem.