FarmX

FarmX is a Streamlit app for crop planning. It combines yield estimation with soil nutrient predictions so growers can compare crops, estimate projected yield, and inspect model-backed nitrogen, phosphorus, and potassium recommendations.

Features

• Projected yield calculator for 22 supported crops
• Side-by-side crop yield comparison
• Model-backed N, P, and K category predictions
• Predicted nitrogen value in kg/ha
• Curated soil testing and agricultural resource links

Repository Layout

.
├── app.py                 # Streamlit entry point
├── cost.py                # Yield estimation logic
├── functions.py           # Model training and prediction helpers
├── farmx/                 # Shared app metadata
├── data/                  # Training dataset
├── models/                # Committed model/scaler artifacts used by Streamlit
├── assets/                # Logo and favicon assets
└── train-models.py        # Regenerates model pickle files from the dataset

The models/*.pkl files are intentionally committed because the Streamlit app loads them at runtime.

Getting Started

Create the locked environment and install its dependencies:

uv sync

Run the app:

uv run streamlit run app.py

The app will be available at http://localhost:8501.

Regenerating Models

If the dataset or feature engineering changes, regenerate the model artifacts:

uv run python train-models.py

This updates the pickle files in models/.

Dependency Notes

Dependencies are declared in pyproject.toml and reproducibly pinned in uv.lock. requests and streamlit are bounded above the vulnerable ranges reported by Dependabot.