Soil Intelligence Crop Recommendation Models

This repository contains pre-trained machine learning models for crop recommendation based on soil and environmental factors. The models are trained using the Crop Recommendation Dataset, which is included in this repository.

Dataset

The dataset (Crop_recommendation.csv) consists of the following features:

• N: Nitrogen content in the soil
• P: Phosphorus content in the soil
• K: Potassium content in the soil
• temperature: Temperature in degrees Celsius
• humidity: Relative humidity percentage
• ph: pH value of the soil
• rainfall: Annual rainfall in mm
• label: Recommended crop (target variable)

Models

The following machine learning models have been trained and saved in the models/ and Advance_models/ directories:

Basic Models (models/)

Model Name	Description
DecisionTree.pkl	Decision Tree Classifier
NBClassifier.pkl	Naïve Bayes Classifier
RandomForest.pkl	Random Forest Classifier
XGBoost.pkl	XGBoost Classifier
label_map.pkl	Label encoding mapping

Advanced Models (Advance_models/)

Model Name	Description
DecisionTree.pkl	Advanced Decision Tree Classifier
NBClassifier.pkl	Improved Naïve Bayes Classifier
RandomForest.pkl	Optimized Random Forest Classifier
XGBoost.pkl	Fine-Tuned XGBoost Classifier
Ensemble.pkl	Ensemble Model (combination of classifiers)
label_map.pkl	Label encoding mapping

Model Performance

Detailed model performance metrics, including confusion matrices and feature importance, are available in the Advance_models/scores_img/ directory.

Usage

These models can be loaded and used for inference in Python using the joblib library:

import joblib
import pandas as pd

# Load model
model = joblib.load('Advance_models/RandomForest.pkl')

# Load label mapping
label_map = joblib.load('Advance_models/label_map.pkl')
reverse_label_map = {v: k for k, v in label_map.items()}

# Example input (replace with real data)
sample_input = pd.DataFrame([[90, 42, 43, 20.87, 82.02, 6.5, 202.93]],
                            columns=['N', 'P', 'K', 'temperature', 'humidity', 'ph', 'rainfall'])

# Predict
prediction = model.predict(sample_input)
predicted_crop = reverse_label_map[prediction[0]]
print("Recommended Crop:", predicted_crop)

Repository Structure

Soil-Intel-Crop-Models/
│── models/
│   ├── DecisionTree.pkl
│   ├── NBClassifier.pkl
│   ├── RandomForest.pkl
│   ├── XGBoost.pkl
│   ├── label_map.pkl
│── Advance_models/
│   ├── DecisionTree.pkl
│   ├── NBClassifier.pkl
│   ├── RandomForest.pkl
│   ├── XGBoost.pkl
│   ├── Ensemble.pkl
│   ├── label_map.pkl
│   ├── scores_img/
│       ├── DecisionTree_confusion_matrix.png
│       ├── DecisionTree_feature_importance.png
│       ├── model_comparison.png
│       ├── NBClassifier_confusion_matrix.png
│       ├── RandomForest_confusion_matrix.png
│       ├── RandomForest_feature_importance.png
│       ├── XGBoost_confusion_matrix.png
│       ├── XGBoost_feature_importance.png
│── Crop_recommendation.csv
│── README.md
│── LICENSE

License

This repository is open for educational and research purposes. Feel free to use and modify it as needed.

Contributors

• Ben - GitHub
• BrianOverkill - GitHub

Acknowledgment

Dataset Source: Crop Recommendation Dataset