Project case study

Fall Foliage

Applied machine learning

A machine learning app that estimates fall-color timing from weather and daylight inputs.

The problem

Peak foliage changes with local conditions. A useful planner needs to translate weather signals into clear seasonal windows.

I connected location-specific scikit-learn models to a Flask interface using temperature, precipitation, and daylight inputs.

Each supported location has a saved model and daylight groups. The app assembles the feature columns expected by that model.

The prediction layer converts model output into nine seasonal windows and keeps the three foliage stages in order.

Flask serves the forms and results, pandas structures inputs, joblib loads models, and NumPy applies output constraints.

The project demonstrates the implementation and its 17-location scope. It does not claim independent forecast accuracy or current conditions.

The app supports 17 U.S. locations and returns early-, mid-, or late-month windows for three foliage stages from September through November.