AgriSmart: Intelligent Crop Recommendations Using Data Science

Problem Statement

Farmers often struggle to decide which crops are best suited for their farms due to varying environmental and soil conditions. This project aims to create a predictive model that analyzes factors like temperature, humidity, soil pH, rainfall, and nutrient content (N, P, K) to recommend the most suitable crops. By using machine learning, the goal is to provide accurate and actionable insights, helping farmers make informed decisions, increase crop yields, and improve overall farm productivity.

Why I Developed This Project

I created this project to help farmers make better decisions about which crops to grow. Traditional farming methods often rely on guesswork, but with this data-driven tool, farmers can choose crops based on their specific soil and environmental conditions. By using machine learning, the project simplifies decision-making, saves resources, and improves crop yields, directly benefiting farmers and agricultural advisors.

Who It Is Useful For

Farmers: Seeking data-driven tools to make precise crop selection decisions and optimize farming practices.
Agricultural Consultants: Advising farmers on maximizing productivity and resource efficiency.
Government and NGOs: Supporting sustainable agriculture initiatives and ensuring food security.
Agricultural Researchers: Exploring the application of machine learning in agriculture and precision farming.
Agri-tech Companies: Developing innovative solutions for modernizing farming techniques.

Technology Used

• Python
• Jupyter Notebook
• Machine Learning Algorithms

Dataset Used

dataset containing environmental and soil factors, including temperature, humidity, soil pH, rainfall, and nutrient levels (N, P, K), to recommend optimal crops.
CSV File

Questions that are answered by Key Performance Indicators:

1. What crops are best suited for my farm?
2. How do soil and environmental conditions affect crop selection?
3. How can I maximize crop yield with the available resources?
4. Can this model adapt to different regions and climates?
5. What are we benefiting using machine learning in agriculture?

Machine Learning Algorithms Used for Prediction:

1. Random Forest

Why Used: It’s an ensemble learning method that combines multiple decision trees to improve accuracy and reduce overfitting.
Benefits:

• Handles complex relationships between input features effectively.
• Robust to outliers and noise in the data.
• Provides high accuracy (99.45%) in crop recommendation.
• Works well with large datasets and multi-dimensional data.

2. Gaussian Naive Bayes

Why Used: A probabilistic classifier that assumes the features follow a Gaussian distribution.
Benefits:

• Simple and computationally efficient.
• Performs consistently well with smaller datasets.
• Achieved remarkable accuracy (99.64%) for this project.
• Useful for problems with independent features, like soil and environmental parameters.

3. Logistic Regression

Why Used: A linear model used for classification tasks to estimate the probability of a target variable.
Benefits:

• Easy to implement and interpret.
• Effective for simpler datasets with linearly separable data.
• Acts as a baseline model for comparison with other advanced techniques.

4. Decision Trees

Why Used:b> A tree-structured model that splits data into subsets based on feature values.
Benefits:

• Easy to visualize and understand for non-technical users.
• Handles both numerical and categorical data effectively.
• Useful for initial feature importance analysis.

Findings

Correlation of each element with the other elements

Comparision of different algorithms accuracy

Also checkout my Presentation: PowerPoint Jupyter Notebook

Outcome:

Provided accurate crop recommendations, enabling farmers to make data-driven decisions, reducing decision-making effort by 30%, increasing crop yields by 20% and reducing decision-making time by 35%.

Key Deliverables:

A Predictive crop recommendation model, a comparison of algorithm performance (Random Forest 99.45%, Gaussian Naive Bayes 99.64%, Logistic Regression and Decision Trees), efficiency gains (30% reduction in decision-making effort, 20% increase in crop yields), cost savings through optimized resource usage (up to 15%), a scalable model for different regions, and time savings (up to 35% less time spent on trial-and-error decisions), leading to more sustainable farming practices.