Crop Recommendation Using Machine Learning Based on Soil, Weather, and Other Agronomic Factors

Abstract

Agricultureremainsacriticalsectorforfood production and rural livelihood across the globe. However, climate change, unpredictable rainfall patterns, declining soil fertility, and improper crop selection often lead to reduced agricultural productivity. Traditionally, farmers rely on experience or local knowledge for crop selection, which may not always result in optimal yield. With the advent of Artificial Intelligence (AI) and Machine Learning (ML), there is a significant opportunity to revolutionize agriculture by providing intelligent crop recommendation systems. This research focuses on developing a machinelearning-based crop recommendation model that utilizes key parameterssuchassoilnutrientcomposition (Nitrogen, Phosphorus, Potassium), temperature, humidity, pH level, and rainfall. Various supervised learning algorithms were employed, including Decision Trees, Random Forests, and Support Vector Machines, to predict the most suitable crop for a given set of environmental and soil conditions. The Random Forest algorithm emerged as the most accurate with an average prediction accuracy of over 96%.

Introduction

Problem Overview:
Agriculture in developing countries suffers from low productivity due to poor crop selection, limited access to data, and outdated farming practices. Farmers often choose crops based on tradition rather than soil and climate suitability, resulting in low yields and soil degradation. A data-driven approach using Machine Learning (ML) can optimize crop selection and promote sustainable agriculture.

Objective:

To develop a ML-based crop recommendation system that suggests the most suitable crop for a region based on soil nutrients and climate data, thereby enhancing yield, profitability, and resource efficiency.

Literature Review Highlights:

• Patil & Kumar (2017): Used Decision Trees based on NPK and rainfall; lacked environmental inputs.

• Singh et al. (2019): Applied Random Forest with improved accuracy using soil + weather data.

• Sharma et al. (2020): Combined SVM and KNN; addressed model overfitting issues.

• Jha et al. (2021): Used deep learning with satellite imagery; high accuracy but costly.

• Kaggle Dataset (2023): Benchmark dataset used widely in agricultural ML applications.

Methodology:

A. Data Sources:

• Soil data from Soil Health Card Scheme

• Weather data from Indian Meteorological Department (IMD)

• Crop data from Kaggle (2023)

Key Features:

• Soil: Nitrogen (N), Phosphorus (P), Potassium (K), pH

• Climate: Temperature, Humidity, Rainfall

B. Preprocessing Steps:

• Handling missing data

• Min-max normalization

• Label encoding for crops

C. Algorithms Tested:

1. Decision Tree

2. Support Vector Machine (SVM)

3. K-Nearest Neighbors (KNN)

4. Random Forest (best performance)

D. Model Training & Evaluation:

• 80/20 train-test split

• 10-fold cross-validation

• Metrics: Accuracy, Precision, Recall, F1-Score

E. Performance Results:

F. System Deployment:

• Developed as a web/mobile app

• Built using HTML, CSS, JavaScript (frontend)

• Backend with Python Flask/Django + scikit-learn

• Accepts soil and climate inputs → returns crop suggestions with fertilizer/water advice

Conclusion

The research successfully demonstrates the potentialofmachinelearninginrecommending optimal crops based on soil and weather parameters. The Random Forest model outperforms others in terms of accuracy, robustness, and generalization. By integrating this model into a user-friendly application, farmerscanreceivereal-time,location-specific crop recommendations. Such tools have the potential to improve agricultural productivity, reduce input waste, and promote sustainable farming. Future enhancements include real- time weather integration, region- specific tuning,multi-languagesupport,andintegration with market demand data to optimize profitability. This research demonstrates the potential of machine learning in transforming traditional agriculture through intelligent crop recommendations.Byanalyzingacombination ofsoilproperties,weatherdata,andagronomic features, the system provides farmers with scientifically backed suggestions that can enhance yield and sustainability. Future work will involve real- time data integration, localization of recommendations based on region, and multi- language interfaces for broader accessibility.

References

[1] Patil, S., & Kumar, R. (2015). Prediction of suitable crops using decision tree in India. International Journal of Computer Applications, 162(11), 12-16. [2] Singh, A., Gupta, R., & Sharma, S. (2019). Precision agriculture using machine learning for crop recommendation. Journal of AgriculturalInformatics,10(2),45-51. [3] Sharma,P.,Verma,D.,&Rathore,N. (2020). Smart farming using SVM [4] Patil, S. S., & Kumaraswamy, Y. S. (2017).\"Agricultural Crop Yield Prediction Using Artificial Neural Network Approach.\"International Journal of Innovative Research in Electrical, Electronics, Instrumentation, and Control Engineering, 5(1), 1–5. [5] Waleed, A., & Khan, Z. (2019). \"Crop Recommendation System Using Machine Learning Algorithms.\"InternationalJournalofScientific& TechnologyResearch,8(10),3000–3004. [6] Chauhan, A., &Khanna, P. (2020). \"Crop Prediction Using Machine Learning Techniques.\" International Research Journal of EngineeringandTechnology(IRJET), 7(4), 4062–4066. [7] KaggleDataset:\"Crop RecommendationDataset.\" Availableat:https://www.kaggle.com/datasets/atharvaingle/crop-recommendation- dataset [8] Bhargava, S., &Pawar, P. (2021). \"SmartFarming:Crop Recommendation System Using MachineLearning.\" International Journal of Advanced ResearchinComputerandCommunicationEngineering,10(4), 1–5. [9] Gandhi, R., &Natarajan, S. (2020). \"Machine Learning Based Predictive Model for Crop Recommendation.\" International Journal of Recent Technology and Engineering (IJRTE), 8(6), 317–320. [10] Liakos,K.G.,etal.(2018). \"Machine learning in agriculture: A review.\"Sensors,18(8),2674.https://doi.org/10.3390/s18082674

Copyright

Copyright © 2025 Prof. S. S. Ganorkar, Nikhil Deshmukh , Ankit Bihone , Prajwal Chandankhede, Nikhil Tembhurne. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.