Best Crop Yield Prediction Using Machine Learning

Crop Yield Prediction

Being able to predict crop yield properly is super important for farmers, agriculture experts, and even the government so they can plan things better. This project uses Machine Learning to help predict how much crop can be produced, based on things like soil, climate, and region. It’s built as a full web app where everything from data to predictions works smoothly in one place.

Overview:

This web app is made to guess the yield of different crops using proper agricultural data. It’s built in a way that you can scale it up, and the design makes it easy to use. The main parts of the system are:

• Data Ingestion
• Data Transformation
• Model Training
• Prediction Pipeline

All of it’s done in Python, and the files are nicely organized to make it easier to manage and grow the project over time.

Project Summary

Download New Real Time Projects :-Click here

Project Structure

• src/: Core logic including model training and prediction code.
• components/: Modular scripts for data ingestion, transformation, and training.
• pipeline/: Training and prediction execution pipelines.
• logs/: Stores logs generated during execution.
• utils/: Utility functions for logging, error handling, and MongoDB integration.
• exception/: Custom exception management.
• app.py: Flask app for deploying the web interface.

Installation

To install the dependencies required for running the project, use the following command:

pip install -r requirements.txt

Required Packages:

• numpy==1.26.4
• pandas==2.2.2
• flask==3.0.3
• scikit-learn==1.4.2
• seaborn==0.13.2
• pymongo==4.6.3
• datetime==5.5

Pipeline Overview

1. Data Ingestion

• Source: MongoDB database.
• Features: Includes Nitrogen (N), Phosphorus (P), Potassium (K), pH, rainfall, temperature, area (in hectares), crop type, crop name, and state.
• Process:
  • Connect to MongoDB and retrieve data.
  • Convert data to CSV for further use.
  • Split into training and testing datasets.

2. Data Transformation

• Imputation:
  • Median for numerical features.
  • Most frequent for categorical features.
• Scaling:
  • StandardScaler applied to all numeric values.
• Encoding:
  • OrdinalEncoder for categorical data.
• Pipeline Management:
  • Managed using ColumnTransformer for modularity.

3. Model Training

• Multiple regression algorithms are tested to find the best-performing model:
  • Linear Regression
  • Ridge Regression
  • Lasso Regression
  • ElasticNet Regression
  • Decision Tree Regression
  • Random Forest Regression
• Models are evaluated using R-squared metric.

4. Prediction Pipeline

• CustomData Class: Accepts input from users and converts it to the correct DataFrame format.
• PredictPipeline Class: Loads saved model and preprocessing pipeline for generating predictions.

Available Features

This project includes all essential components needed for accurate crop yield prediction:

• Dynamic web-based user input form
• Integrated prediction engine with machine learning models
• MongoDB data integration
• Exception and logging system
• Modular pipeline architecture (training and prediction separated)
• Real-time yield prediction output based on input data

Use Case

This application can be used by:

• Agricultural researchers for data analysis and prediction.
• Farmers and agro-business consultants for planning and optimizing crop output.
• Policy makers for estimating food production and planning logistics.

We have projects Available in all languages:–Click Here