Crop Recommendation System Based on Soil Analysis 🌱

Project Overview

Soil analysis plays a crucial role in determining crop productivity. This project utilizes soil properties such as pH, nutrient levels (N, P, K), organic matter percentage, and humidity to predict the optimal crop using a Random Forest Classifier. Regional suitability for crops across Turkey's seven geographical regions is included to narrow down recommendations effectively.

---

Features

• Input Parameters: Soil pH, Nitrogen, Phosphorus, Potassium, Organic Matter, and Humidity.
• Regional Suitability: Incorporates regional data (Marmara, Aegean, Mediterranean, Central Anatolia, Eastern Anatolia, Black Sea, and Southeastern Anatolia).
• Machine Learning Model: Optimized Random Forest Classifier with hyperparameter tuning.
• Prediction Capability: Provides crop predictions based on soil input and region-specific suitability.
• User-Friendly Dataset: Clear labeling and well-organized data structure for ease of use.

---

Dataset

The dataset includes:

1. Soil Analysis Columns:

• PH: Soil pH value
• N(mg/kg): Nitrogen content
• P(mg/kg): Phosphorus content
• K(mg/kg): Potassium content
• ORG(%): Organic matter percentage
• HUM(%): Humidity percentage

2. Target Variable:

• CROP: Crop name (e.g., bugday, arpa, misir, etc.)

3. Region Columns:

• REGION_MARMARA, REGION_AEGEA, REGION_MEDITERRANEAN, etc.
  Binary flags (0 or 1) indicating whether the crop is suitable for the corresponding region.

---

Technologies Used

• Python
• Scikit-learn (Machine Learning)
• Pandas and NumPy (Data Preprocessing)
• Matplotlib (Visualization)
• Git and GitHub (Version Control)

---

Model Details

• Algorithm: Random Forest Classifier
• Hyperparameters:
  • n_estimators: 322
  • max_depth: 14
  • max_features: 'log2'
  • min_samples_split: 5
  • min_samples_leaf: 2
  • bootstrap: False
• Evaluation Metrics:
  • Accuracy on Test Data
  • Cross-Validation Scores

---

Installation

Prerequisites

Ensure you have the following installed:

• Python 3.8+
• Git
• Libraries: pandas, scikit-learn, numpy, matplotlib

Steps

1. Clone the repository:

   git clone https://github.com/your-organization/crop-recommendation-system.git
   cd crop-recommendation-system

2. Install dependencies:

    pip install -r requirements.txt

3. Run the project scripts:

    python app.py
   
   

Usage

Prepare Input Data:

• Provide soil parameters (pH, N, P, K, ORG, HUM) and regional information.

Run the Model:

• Use the trained model to predict suitable crops.

Example Prediction:

new_sample = [[6.2, 33, 28, 290, 2.9, 71]]  # Soil properties
predicted_label = model.predict(new_sample_df)
print("Recommended Crop: ", label_encoder.inverse_transform(predicted_label)[0])

CONTACT:

For any questions or issues, please contact ozgulbaytekin or sedefelms.