FarmEasy

This project provides a machine learning-based crop recommendation system. The system uses environmental parameters such as nitrogen, phosphorus, potassium levels, temperature, humidity, pH, and rainfall to predict the most suitable crop for a given set of conditions.

Folder Structure

Crop-Recommendation-System/
│
├── dataset/
│   └── crop_recommendation.csv
│
├── images/
│   └── crops/
│       └── <crop_images>.jpg
│
├── models/
│   └── model_files.pkl
│
├── nav/
│   ├── home.py
│   ├── predict.py
│   └── visualize.py
│
├── app.py
├── train_model.py
├── database.py
├── requirements.txt
└── README.md

CREATE Virtual Environment and Install Requirements

First, create a virtual environment to manage dependencies for this project.

virtualenv venv
source venv/bin/activate
pip install -r requirements.txt

Run Application

To run the application, ensure your virtual environment is active, then execute the following commands:

source venv/bin/activate
python3 train_model.py
streamlit run app.py

Files and Directories

dataset/: Contains the crop_recommendation.csv dataset.
images/crops/: Contains images of the recommended crops.
models/: Directory where model files are stored after training.
nav/: Contains navigation modules for the Streamlit application.
- home.py: Home page of the application.
- predict.py: Prediction page for recommending crops.
- visualize.py: Visualization page for displaying model metrics and visualizations.
app.py: Main entry point for the Streamlit application.
train_model.py: Script for training and saving machine learning models.
database.py: Handles logging predictions to the database.
requirements.txt: Lists required Python packages for the project.
README.md: Project documentation file.

Additional Information

Model Training: The train_model.py script trains three models (Logistic Regression, Decision Tree, and Random Forest) and saves the trained models, metrics, and label encoder in the models/ directory.
- Logistic Regression: A linear model for binary classification.
- Decision Tree: A non-linear model that splits the data based on feature values.
- Random Forest: An ensemble model that combines multiple decision trees to improve accuracy and control overfitting.
Model Prediction: The predict.py script uses the trained models to recommend the most suitable crop based on user input parameters.
Visualization: The visualize.py script provides visualizations for the model metrics and decision tree.

Model Metrics

The following metrics are used to evaluate the performance of the models:

Accuracy: The ratio of correctly predicted crops to the total crops.
Precision: The ratio of correctly predicted positive observations to the total predicted positive observations.
R^2 Score: The proportion of the variance in the dependent variable that is predictable from the independent variables.

After training the models, the metrics are as follows:

Logistic Regression
- Accuracy: 95.42%
- Precision: 96.04%
- R^2 Score: 89.67
Decision Tree
- Accuracy: 98.63%
- Precision: 98.71%
- R^2 Score: 96.20
Random Forest
- Accuracy: 99.24%
- Precision: 99.34%
- R^2 Score: 97.3

ashuthe1 / FarmEasy