GoviCare Crop Risk Detection Model

An ensemble classifier (RandomForest + GradientBoosting + ExtraTrees) that predicts whether a crop is at risk based on environmental sensor data.

Performance

Metric	Score
Accuracy	0.9004
F1 Score	0.8834
Precision	0.9809
Recall	0.8035
ROC AUC	0.9676

Supported Crops (25 total)

banana, barley, cabbage, carrot, chili_pepper, cinnamon, coconut, corn, cotton, cucumber, eggplant, groundnut, lettuce, mango, onion, papaya, potato, rice, rubber, sorghum, soybean, sugarcane, tea, tomato, wheat

Input Features

• crop_type (string) - one of the supported crops
• soil_moisture (float, 0-100%)
• temperature (float, Celsius)
• humidity (float, 0-100%)
• wind_speed (float, m/s)
• pressure (float, hPa)

Usage

import joblib
from huggingface_hub import hf_hub_download

model = joblib.load(hf_hub_download("dimeshanthoney/govicare", "crop_risk_model.joblib"))
scaler = joblib.load(hf_hub_download("dimeshanthoney/govicare", "scaler.joblib"))
encoder = joblib.load(hf_hub_download("dimeshanthoney/govicare", "encoder.joblib"))

Training

Trained on 125,000 synthetic samples across 25 crop types using scientifically-backed agronomic thresholds from FAO, IRRI, and USDA sources.