TesnorFlow Fertilizer and Crop Recommendation system

GeziraPal is a project that aims at assisting individuals with knowledge sharing in agriculture & crop cultivation. The project utilizes TensorFlow, a popular machine learning framework, to develop a recommendation systems for suggesting suitable crops to cultivate based on various factors (atomeshperic parameters and soil readings). so far deployed in Streamlit. watching this demo video for more information.

How to Use the Code

Loading and Installing Dependencies

Before running the application, ensure you have installed the necessary dependencies. You can find all the required dependencies along with their corresponding versions in the requirements.txt file. we recommend using github virtual enviroment using the command pip install -r requirements.txt you can install all the software dependcies at once

import numpy as np
import pandas as pd
import joblib
import pickle
from sklearn import *
import tensorflow as tf
import tensorflow.keras as keras
import streamlit as st

To use GeziraPal locally, follow these steps:

1/ Clone the repository to your local machine using the following command:

git clone https://github.com/A7med7x7/GeziraPal.git

2/ Install the necessary dependencies listed in the requirements.txt file using:

pip install -r requirements.txt

3/ Run the Streamlit web application by executing:

streamlit run app.py through bash/zsh CLI

4/ you're ready to go!

Usage Details

NumPy & Pandas: Use these libraries for numerical computations and data manipulation within your code. Joblib: Use it as an alternative for loading models, encoders, and scalers saved during preprocessing and model training. Scikit learn: is used for machine learning algorithms but mainly used for scoring the models from metrics class. TensorFlow: for machine learning model development / NN architecture. Streamlit: Utilized for streamlining the web application development process.

About the Data

Source

The dataset was originally obtained from Kaggle: Crop and Fertilizer Dataset for Western Maharashtra. under the name 'Crop and fertilizer dataset.csv' in the data directory

Overview

• The dataset is generated and contains 11 features with more than 4k instances.
• After analysis and data generation, it now has over 12k instances.
• The target variable is the crop type.
• In this repository, you will have access to the cleaned version of the dataset.
  1. Crop and fertilizer dataset.csv: the raw dataset
  2. crop_and_fertilizer.csv: The full dataset.
  3. x_train_resampled.csv: The training variables.
  4. y_train_resampled.csv: Target variable.

Notebook Contents

The notebook consists of:

1. Data reading and cleaning.
2. Data preprocessing.
3. Modeling.
4. Evaluation and systemizing functions.

String Value Input Function

In order to enable users to input string values instead of labeled numbers to represent their information, a function is built using transformers (scaler and encoder).

Prediction Functions in the App

The app contains 3 prediction functions:

1. Basic Recommendation Function:

   • It's a filtering method that uses only Pandas as a tool without any prediction capabilities or modeling.
   • It returns the available crops to grow from the database that match your inputs.

2. Advanced Recommendation Function:

   • Using TensorFlow, we build a multi-class classification Keras model that is able to recommend the best crop.

3. Corresponding YouTube Tutorial Function:

   • This function returns the predicted label, converts it, and then searches in the corresponding YouTube tutorial from the 'link' column to suggest what to watch.