---
license: apache-2.0
tags:
  - tabular-classification
  - fertilizer-recommendation
  - agriculture
  - scikit-learn
library_name: scikit-learn
---

# AgroMind Fertilizer Prediction Model

## Model Description
Scikit-learn classifier that predicts the most suitable fertilizer based on soil conditions, crop type, and environmental factors.

## Framework
- **Library**: scikit-learn
- **Format**: pickle (.pkl)
- **Includes**: classifier model + LabelEncoder for fertilizer names

## Input Features
| Feature      | Type        |
|--------------|-------------|
| Temperature  | int (0–100) |
| Humidity     | int (0–100) |
| Moisture     | int (0–100) |
| Soil Type    | encoded int (Black=0, Clayey=1, Loamy=2, Red=3, Sandy=4) |
| Crop Type    | encoded int (Barley=0, Cotton=1, … Wheat=10) |
| Nitrogen     | int (0–100) |
| Potassium    | int (0–100) |
| Phosphorus   | int (0–100) |

## Usage
```python
from huggingface_hub import hf_hub_download
import pickle, numpy as np

repo = "Arko007/agromind-fertilizer-prediction"
with open(hf_hub_download(repo, "classifier.pkl"), "rb") as f:
    clf = pickle.load(f)
with open(hf_hub_download(repo, "fertilizer.pkl"), "rb") as f:
    le = pickle.load(f)

features = np.array([[28, 65, 40, 2, 6, 50, 40, 30]])  # [temp, hum, mois, soil, crop, N, K, P]
pred_idx = clf.predict(features)
fertilizer = le.inverse_transform(pred_idx)
```

## Output
Fertilizer name (string) via LabelEncoder inverse transform.