Machine learning-based optimal crop selection system in smart agriculture.

Area of Science:

• Agricultural Science
• Data Science
• Environmental Science

Background:

• Crop cultivation is highly dependent on regional weather patterns, making agro-climatic analysis crucial for maximizing yield.
• Selecting the right crop for specific land and season is vital for agricultural productivity.
• Machine learning offers advanced solutions for analyzing complex environmental data to aid agricultural decision-making.

Purpose of the Study:

• To develop and evaluate a machine learning-based model for optimal crop selection.
• To integrate weather condition analysis and soil parameters for improved agricultural planning.
• To enhance crop yield prediction through accurate weather forecasting and timely sowing recommendations.

Main Methods:

• Utilized Long Short-Term Memory Recurrent Neural Networks (LSTM RNN) for precise weather condition prediction (temperature and rainfall).
• Employed Random Forest Classifier for accurate crop selection, resource dependency assessment, and optimal sowing time determination.
• Integrated weather data analysis with soil parameters to create a comprehensive decision-making model.

Main Results:

• LSTM RNN achieved low Root Mean Square Error (RMSE) in weather predictions: 5.023% for Min. Temp., 7.28% for Max. Temp., and 8.24% for Rainfall.
• Random Forest Classifier demonstrated high accuracy: 97.235% for crop selection, 96.437% for resource dependency, and 97.647% for sowing time.
• The model construction time was efficient at 5.34 seconds for the Random Forest Classifier.

Conclusions:

• The proposed ML model effectively integrates weather and soil data for informed crop selection and agricultural planning.
• LSTM RNN and Random Forest Classifier provide accurate and efficient tools for agro-climatic analysis and yield optimization.
• The model shows significant potential for improving agricultural practices and increasing crop yields through data-driven insights.