You can deploy this model and start earning money today!

Fatima-Z/
crop-yield-prediction-model

Name: crop-yield-prediction-model
Availability: InStock

joblib

agriculture

crop-yield-prediction

Views

No views yet

🌾 Optimized Crop Yield Prediction AI System

An end-to-end Machine Learning web application designed to forecast agricultural crop yields, recommend optimal crops based on regional climate parameters, and simulate environmental sensitivity curves (Rainfall, Temperature, and Pesticide application).

🌟 Key Features

🔮 Interactive Crop Yield Forecaster:
- Supports 101 Countries and 10 Major Staple Crops.
- Input Climate metrics: Annual Rainfall (mm), Average Temperature (°C), Pesticides (Tonnes), and Year.
- Calculates yield in Tonnes/Hectare (t/ha), hg/ha, kg/ha, kg/acre, and Total Farm Harvest.
- Agricultural Yield Performance Rating Badge (🌟 Exceptional, 🟢 Above Average, 🟡 Normal, 🔴 Low) with historical comparisons.
- ⚡ 1-Click Country Defaults: Automatically loads historical average climate metrics for any selected country.
🌾 Multi-Crop Optimization & Recommender:
- Evaluates all 10 major crops simultaneously under the same climate conditions.
- Highlights the #1 Highest Yielding Crop with interactive Plotly ranking charts.
⚙️ Climate & Input Sensitivity Simulator:
- Interactive "What-If" response curves:
  - Rainfall vs Yield: Identifies dry vs optimal vs flood zones.
  - Temperature vs Yield: Pinpoints thermal growth sweet spots.
  - Pesticide vs Yield: Illustrates application efficacy and diminishing return plateaus.
📊 Model Benchmarking & Feature Insights:
- Compares 5 ML algorithms (Extra Trees, Random Forest, Decision Tree, LightGBM, XGBoost).
- Grouped and individual feature importance charts.
📁 Dataset Explorer & Climate Insights:
- Interactive data explorer previewing historical records and summary statistics.
🚀 REST & Python API:
- Built-in gradio_client and REST endpoints for programmatic agricultural forecasting.

📊 Dataset & Model Architecture

Dataset Overview (Kaggle / FAO)

Total Records: 28,242 historical agricultural observations (1990 – 2013).
Geographic Coverage: 101 countries worldwide.
Crops (10): Cassava, Maize, Plantains and others, Potatoes, Rice (paddy), Sorghum, Soybeans, Sweet potatoes, Wheat, Yams.
Features: Area, Item, Year, average_rain_fall_mm_per_year, pesticides_tonnes, avg_temp.
Target Variable: hg/ha_yield (Hectograms per Hectare).

Benchmark Performance

Model Algorithm	R² Score	MAE (hg/ha)	RMSE (hg/ha)	Training Time
Extra Trees (Champion)	0.9913 (99.13%)	2,570.21	7,925.48	~3.1s
Random Forest	0.9876 (98.76%)	3,452.75	9,470.36	~2.6s
Decision Tree	0.9794 (97.94%)	3,628.86	12,210.03	~0.5s
LightGBM	0.9679 (96.79%)	8,649.42	15,259.90	~2.4s
XGBoost	0.9621 (96.21%)	9,904.11	16,569.78	~0.7s

🛠️ Local Installation & Running

bash

# 1. Clone the repository
git clone https://github.com/YOUR_USERNAME/optimized-crop-yield-prediction.git
cd optimized-crop-yield-prediction

# 2. Install dependencies
pip install -r requirements.txt

# 3. (Optional) Re-train the model
python train_model.py

# 4. Launch the Gradio Web Application
python app.py

Open your browser at http://localhost:7860.

🚀 Deploying to Hugging Face Spaces

Create a new Space at huggingface.co/new-space.
Select Gradio as the Space SDK.
Upload the following files to your Space:
- app.py
- requirements.txt
- crop_yield_model.joblib
- model_metadata.json
- yield_df.csv
- README.md
Hugging Face will automatically build the environment and start the app in ~60 seconds!

Detailed instructions are available in DEPLOYMENT_GUIDE.md.

📜 License

This project is licensed under the MIT License.

Fatima-Z/crop-yield-prediction-model