Crop & Fertilizer Recommendation System.

Problem Statement

Farmers often rely on traditional knowledge or guesswork to decide which crop to plant and which fertilizer to use. This leads to reduced yield, inefficient fertilizer use, and higher costs. The challenge was to build a data-driven decision support system that could guide farmers with accurate predictions in real time.

Technical Solution

1. Step 1 – Architecture & Approach:
   • Built a Flask-based web application where soil and weather data are collected as inputs.
   • Data undergoes preprocessing — including feature scaling and handling missing values.
   • The cleaned data is passed into a Random Forest Classifier model.
   • The model predicts the optimal crop type and recommends a matching fertilizer.
2. Step 2 – Algorithms, Libraries & Highlights:
   • Machine Learning: Random Forest Classifier for multi-class crop prediction.
   • Libraries: Pandas (data handling), scikit-learn (model building & evaluation), NumPy (mathematical operations).
   • Interface: Tkinter GUI for desktop version and Flask backend for web deployment.
   • Highlights: Achieved high prediction accuracy and reduced overfitting using ensemble learning.
3. Step 3 – Deployment & Infrastructure:
   • Backend hosted on Render Cloud for reliability and scalability.
   • Packaged ML model integrated with Flask routes for real-time predictions.
   • Future scalability planned using AWS SageMaker and mobile app integration.

Results & Impact

Enabled data-driven crop selection, reducing dependency on guesswork.

Technologies Used