Educational Administration: Theory and Practice

2024, 30(5), 11035-11041

ISSN:2148-2403
https://kuey.net/ Research Article

Agricultural Crop Analysis using IoT And Machine

Learning
Akhil Panwar1*, Arpit Kumar2, Aryan Poonia3, Ashish Chauhan4, Ruchin Taliyan5, Vijay kumar Sharma6
1*,2,3,4,5,6Department
of CSE, MIET, Meerut.
Email:-1*[email protected],[email protected],[email protected]
[email protected], [email protected], [email protected]

Citation: Akhil Panwar, et al (2024), Agricultural Crop Analysis using IoT And Machine Learning, Educational Administration: Theory
and Practice, 30(5), 11035-11041
Doi: 10.53555/kuey.v30i5.4881

ARTICLE INFO ABSTRACT

The power of the IoT and machine learning could radically transform agriculture
using precision agriculture. Sensor data in real-time is available to farmers,
allowing them to select crops and management techniques in an enlightened
manner. Machine learning algorithms process the information to predict which
crops are best suited for a location’s soil type, weather conditions, and other
factors. This combination leads to increased resource use efficiency, improved
crop production, proactive risk management, more responsible agriculture, and
an opportunity for farmers to obtain data-driven suggestions. To maximize the
benefits of these technologies, it is important to address problems like data
rights and availability, as well as connectivity issues. This would allow for the
continuation of R&D in the field.

Abbreviations: Internet of things(IoT), Research and Development(R&D),

Machine Learning(ML), Crop suitability mapping

1. Introduction:

The predominant challenge in agriculture lies in comprehending the fluctuating climate patterns, crucial for
optimal crop growth. Precision farming offers a solution, sustaining productivity and boosting yields by
aligning farming practices with specific crop requirements. In India, sustaining agriculture is vital amid rising
demands, where traditional methods’ limitations can be mitigated through precision farming[1]. Leveraging
IoT and predictive systems, precision farming aids decision-making by collecting field data for
recommendations. Despite efforts to minimize losses, challenges persist in crop selection and adapting to
climate shifts, exacerbated by shortcomings in current prediction methods. The proposed system aims to
address these issues by enhancing yields, conducting real-time crop analysis through IoT, and optimizing
parameter selection for informed decisions. Effective algorithms are pivotal for early crop prediction, utilizing
ML models to generate valuable data for crop selection. The crop suggestion model’s primary goal is to reduce
losses by recommending suitable crops for specific fields, necessitating careful algorithm selection to ensure
accuracy. Machine Learning emerges as the most promising technology for precise crop prediction and yield
estimation[2].

2. Related Works

Forecasting crop yields is crucial for decision-makers at national and regional levels, expediting informed
decisions. Accurate prediction models empower farmers in critical choices regarding crop selection and
cultivation timing. Various methodologies, particularly machine learning (ML) techniques, have been explored
for crop yield prediction. This section offers an overview of pertinent literature, excluding surveys and
traditional reviews, to illuminate research trends in this area[4].
Chlingaryan and Sukkarieh (2018) conducted a thorough review on nitrogen status estimation using ML
techniques, stressing the importance of sensing technologies and ML advancements in cost-effective
agricultural solutions[5]. Elavarasan et al. (2018) surveyed ML models for crop yield prediction, emphasizing
the significance of incorporating diverse climatic parameters for improved accuracy[6].
Liakos et al. (2018) provided a review encompassing ML applications in various agricultural domains,
highlighting ML’s diverse impact on agricultural practices. Li, Lecourt, and Bishop (2018) examined fruit
Copyright © 2024 by Author/s and Licensed by Kuey. 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.
11036 Akhil Panwar, et al / Kuey, 30(5), 4881

ripeness determination for optimal harvest timing and yield prediction, underscoring the necessity of accurate
predictive models in enhancing agricultural productivity[7].
Mayuri and Priya addressed challenges and methodologies in image processing and ML for agricultural disease
detection, illuminating advancements in disease management. Somvanshi and Mishra (2015) explored ML
approaches in plant biology, demonstrating ML’s applications in understanding and improving plant-related
processes[8].
Gandhi and Armstrong (2016) reviewed data mining applications in agriculture, advocating for further
research to integrate data mining techniques into complex agricultural datasets. Beulah (2019) surveyed data
mining techniques in crop yield prediction, promoting their use in addressing predictive modeling
challenges[9].
Our study presents the first systematic literature review (SLR) solely focusing on ML in crop yield prediction.
Unlike existing surveys, which often focus on specific aspects, our SLR comprehensively reviews the literature
in this field. Additionally, we analyze 30 deep learning-based studies, showcasing the effectiveness of deep
learning algorithms in tackling crop yield prediction challenges.

3. Proposed Work Plan

3.1 General architecture (Figure 1)

The approach is to study the To optimize agricultural analysis through Machine Learning (ML), a methodical
approach is vital. It begins with clearly defining the agricultural problem to be addressed, whether it's
improving crop yield, detecting diseases, or assessing soil health. Next, comprehensive datasets are collected
from various sources such as IoT sensors, satellite imagery, and historical agricultural records. These datasets
undergo preprocessing steps, including data cleaning, normalization, and feature selection, to ensure they are
suitable for ML modeling. Upon successful validation, the optimized ML models are deployed in agricultural
settings to aid farmers in making data-driven decisions. Continuous monitoring and refinement of these
models ensure their efficacy in improving productivity.
Description of various modules of the system.

A. Data Collection
Historical data on many important factors is important for analyzing crop impacts. Climate models covering
temperature changes, precipitation, humidity and wind dynamics form an important part of this analysis.
Tracking these patterns over time can help you understand how different climates affect crop growth and
production. Soil quality is another important consideration and requires careful analysis of historical data on
soil composition, pH, nutrient content and organic matter[16]. It provides information about the health
conditions of the soil and the ability of the soil to support healthy crops. Satellite images help monitor soil
moisture over large agricultural areas. Stakeholders can influence actions by combining information from
historical events with dynamic data provided by remote sensing data, satellite imagery, and ground sensors. It
can be determined that farms will increase productivity and ensure food security. This collaboration reflects
the evolution of technology and data-driven approaches in agriculture today[11].

B. Data Pre-processing
After collecting data on various factors affecting the crop, the next important step is to clean the data and
prepare it for analysis. Data cleaning involves identifying and resolving discrepancies, and missing values that
may affect analysis[12]. This process ensures the reliability and accuracy of the data set and creates a solid
foundation for subsequent analysis. Standardization scales the importance of different variables into a
standard range (usually between 0 and 1) and makes it easier to compare variables with different scales and
units. Standardization changes the data so that the mean is 0 and the standard deviation is 1; This reduces the
impact of outliers and brings the distribution of the data closer to a Gaussian distribution[13]. These
techniques help control the training process of machine learning models and improve their convergence and
efficiency. Structure and relationship. Similarly, precipitation classification can be made by collecting
precipitation data for a certain time or region. Soil nutrient levels can be expressed as a combination or ratio
derived from soil data[14][14]. By extracting and analyzing these important features, the forecast model
becomes more powerful and accurate, allowing for better predictions of agricultural yields and interruptions.
Through careful data management, standardization and feature engineering, the data set is improved and
enriched, thus providing the basis for in-depth analysis and effective prediction of crop yield.

C. Model Analysis
Choosing an appropriate machine learning method for crop forecasting requires careful consideration of many
factors, including model complexity[19], interpretability, and prediction performance. Regression models such
as linear regression or polynomial regression are often used to predict crop yield when there is a relationship
between input and target variables[20]. This model is simple and easy to understand; It makes it easier to
understand how each input affects the estimated yield. However, they may not capture the nonsocial
relationships present in the data, which may limit their effectiveness. Decision trees and random forests are
popular for product yield because they can capture non-linear relationships and interactions between inputs.
 Akhil Panwar, et al / Kuey, 30(5), 4881 11037

Deep learning models, in particular, provide the flexibility to learn complex patterns and relationships from
data, making them powerful for crop prediction. Deep neural networks, such as convolutional neural
networks (CNN) or recurrent neural networks (RNN), enable more complex analysis and predictions[21] by
extracting features from raw data. However, reporting these models often requires extensive data and budget,
and their black status can hinder interpretation compared to simple models. On the other hand, when the data
show no correlations and present complex patterns that simple models cannot capture, more complex models
such as random forests or neural networks[22] would be expected to say yes. Finally, algorithm selection
should be guided by the overall goal of accurate yield prediction while balancing model complexity and
interpretation assumptions.

D. Training of Model
After collecting and processing the dataset for crop prediction, the subsequent stage involves categorizing the
data into training and validation sets to assess the learning model's effectiveness. This partitioning enables
training on a subset of the data and evaluating performance independently, offering an impartial
evaluation[23]. The training set is utilized for model training, while the validation set is employed to assess
performance and fine-tune hyperparameters. Cross-validation entails dividing the data into multiple subsets or
folds, training the model on different folds, and evaluating performance on the remaining folds. This iterative
process enhances reliability by minimizing variation introduced by training-validation splits, ensuring the
model generalizes well to new data[24]. Using the validation set, the selected model is trained,
hyperparameters are adjusted, and cross-validation is employed for comprehensive evaluation and
optimization. This iterative training and testing approach aids in selecting suitable algorithms and metrics for
accurate crop yield prediction.

E. Testing
Once you choose a machine learning model that has been trained using historical data and fine-tuned with
hyperparameter tuning and cross-validation, the next step is to evaluate its performance using appropriate
metrics. path. The model with the lowest MAE and RMSE value and the highest R-squared value is generally
considered the best performing model. However, other factors such as computational efficiency,
interpretability, and robustness should also be taken into account when choosing the final model. By
evaluating and comparing the performance of different models using appropriate metrics, agriculturalists can
make informed decisions and use the best models to make accurate and reliable prediction of crop fulfills the
desired user requirements, and handles all exceptional cases. To test the model's accuracy, we split the labeled
data into 80% of the labeled datasets as training data and 20% as data to be tested. After the application of the
data that is trained and tested with the Naive Bayes algorithm, we get the precision of 85.4% in our
system/study[25].

F. Architecture
The system involves real-time data collection, prediction model creation, and development of a user interface
for input provision. Initially, data preprocessing occurs. Following preprocessing, a machine learning
algorithm generates a prediction model. Test data is then fed into this model for prediction[26]. The model
undergoes testing with random input values to assess accuracy and error during prediction. This process
iterates until error reduction and accuracy enhancement are achieved. To gather inputs from both IoT sensors
and users, a graphical user interface (GUI) is utilized. These inputs are then utilized by the crop prediction
model to accurately forecast crops[27].

(Figure 1)
11038 Akhil Panwar, et al / Kuey, 30(5), 4881

G. Feedback Loop
Creating feedback loops is important to evaluate the effectiveness of decision-making algorithms based on
their impact on actual crops. This feedback involves collecting data on actual crops processed by farmers based
on the algorithm's recommendations. By comparing the actual results with the predicted results and the
recommendations generated by the model[28], the performance of the algorithm can be evaluated. Farmers
can understand the effectiveness of the consensus model and detect discrepancies between predictions and
actual results. Agricultural experts can provide specific knowledge and skills to enhance the model's
understanding of complex agricultural processes and increase its accuracy. By integrating feedback from
multiple stakeholders, the model continues to evolve and adapt to changes, making it effective and efficient in
agricultural decision support. This positive feedback has led to collaborations for model improvements, driving
continuous improvement and innovation in crop forecasting and agricultural management [35].

3. Experimental Result Analysis:

3.1 Description of the data set used.
EXPERIMENTAL RESULTS AND ANALYSIS
This paper utilized an agricultural dataset containing exclusive soil and environmental data, sourced directly
from the farming community due to its unavailability publicly. The study evaluates feature selection and
classification techniques employing metrics such as accuracy, specificity, recall, precision, F1 score, mean
absolute error, log loss, and area under the curve. The results are tabulated in Table 1 to Table 5 [33][34].
Table 1 illustrates that the random forest algorithm achieves the highest accuracy, followed by k-nearest
neighbor and bagging classifiers. Table 2 shows that random forest, when combined with sampling techniques,
effectively predicts crops. Here, different sampling methods are analyzed for dataset balancing. Table 3
explores optimal feature selection techniques across various[29] classifiers using the felin dataset, considering
MRFE, RFE, and Boruta methods. Table 4 evaluates MRFE's performance with RF using fold validation, while
Table 5 assesses MRFE's performance with RF using data splitting validation [31][32].
Tables 4 and 5 demonstrate random forest's performance under different validation methods, showcasing the
evaluation of MRFE and RF techniques. The results suggest that as the range of characteristics expands, the
measured values tend to decrease.
 Akhil Panwar, et al / Kuey, 30(5), 4881 11039

4. Conclusion

The integration of IoT and machine learning in precision agriculture promises transformative benefits. Real-
time sensor data and ML algorithms optimize crop selection and management, boosting resource efficiency
and yields while managing risks sustainably. The study highlights the efficacy of the random forest algorithm,
especially when coupled with feature selection methods like MRFE, for accurate crop yield prediction. It
stresses the importance of thorough data preprocessing for model reliability. However, challenges like data
rights and connectivity issues need resolution for widespread adoption. The feedback loop ensures continuous
model enhancement through real-world data and expert input, fostering collaboration and innovation in crop
management. This research lays a strong foundation for advanced precision agriculture systems, vital for
addressing sustainability and food security demands. It emphasizes ongoing improvements in predictive
models and collaborative efforts among stakeholders. In summary, this study [30]sets the stage for future
research and practical applications driving agricultural transformation through data-driven decision-making
and technological innovation.

5. References:

1. Athani, S., Tejeshwar, C. H., Patil, M. M., Patil, P., & Kulkarni, R. (2017, February 1). Soil moisture
monitoring using IoT enabled arduino sensors with neural networks for improving soil management for
farmers and predict seasonal rainfall for planning future harvest in North Karnataka — India
2. Biradar, H. B., & Shabadi, L. (2017). Review on IOT based multidisciplinary models for smart farming.
3. Mehta, P., Shah, H., Kori, V., Vikani, V., Shukla, S., & Shenoy, M. (2015). Survey of unsupervised machine
learning algorithms on precision agricultural data. IEEE Xplore.
4. Sandhiya, D., & Thiyaneswaran, B. (2017). Extraction of dorsal palm basilic and cephalic hand vein
features for human authentication system. 2017 International Conference on Wireless Communications,
Signal Processing and Networking (WiSPNET). IEEE.
5. Radhika, Y., & Shashi, M. (2009). Atmospheric Temperature Prediction using Support Vector Machines.
11040 Akhil Panwar, et al / Kuey, 30(5), 4881

International Journal of Computer Theory and Engineering, 55–58.

6. Ahmad, S., Kalra, A., & Stephen, H. (2010). Estimating soil moisture using remote sensing data: A
machine learning approach. Advances in Water Resources, 33(1), 69–80.
7. Kumar, R., Singh, M. P., Kumar, P., & Singh, J. P. (2015). Crop Selection Method to maximize crop yield
rate using machine learning technique. IEEE Xplore.
8. Thiyaneswaran, B., Anguraj, K., Kumarganesh, S., & Thangaraj, K. (2020). Early detection of melanoma
images using gray level co‐occurrence matrix features and machine learning techniques for effective
clinical diagnosis. International Journal of Imaging Systems and Technology, (ima.22514).
doi:10.1002/ima.22514
9. Shah, N. P., & Bhatt, P. (2017). Greenhouse Automation And Monitoring System Design And
Implementation. International Journal of Advanced Research in Computer Science, 8(9), 468–471.
10. Jayanthi, J., and J. Selvakumar. "A novel framework to facilitate personalized web search in a dual mode."
Cluster Computing 20.4 (2017): 3527-3535.
11. Parker, C. (2012). Unexpected challenges in large scale machine learning. Proceedings of the 1st
International Workshop on Big Data, Streams and Heterogeneous Source Mining Algorithms, Systems,
Programming Models and Applications - BigMine ’12.
12. B Thiyaneswaran, K Anguraj, M Sindhu, N. S Yoganathan, J Jayanthi. (2020). Development of Iris
Biological Features Extraction for Biometric Based Authentication to Prevent Covid Spread. International
Journal of Advanced Science and Technology, 29(3), 8266–8275.
13. Pudumalar, S., Ramanujam, E., Rajashree, R. H., Kavya, C., Kiruthika, T., & Nisha, J. (2017). Crop
recommendation system for precision agriculture. IEEE Xplore.
14. Anitha, P., & Chakravarthy, T. (2018). Agricultural Crop Yield Prediction using Artificial Neural Network
with Feed Forward Algorithm. International Journal of Computer Sciences and Engineering, 6(11), 178–
181. https://doi.org/10.26438/ijcse/v6i11.178181
15. B Thiyaneswaran, V Bhuvaneshwaran, M.S Dharun, K Gopu, T Gowsikan (Ed.). (2020). Breathing level
monitoring and alerting by using embedded IOT (Vol. 10). Journal of Green Engineering.
16. Rekha, P., Rangan, V. P., Ramesh, M. V., & Nibi, K. V. (2017). High yield groundnut agronomy: An IoT
based precision farming framework. IEEE Xplore.
17. Gupta, A., Nagda, D., Nikhare, P., & Sandbhor, A. (2021). Smart Crop Prediction using IoT and Machine
Learning. International Journal of Engineering Research & Technology, 9(3).
18. Biradar, H. B., & Shabadi, L. (2017). Review on IOT based multidisciplinary models for smart farming.
IEEE Xplore.
19. Jayanthi, J., and J. Selvakumar. "A novel framework to facilitate personalized web search in a dual mode."
Cluster Computing
20. 4 (2017): 3527-3535. 20. Priya, P. K., & Yuvaraj, N. (2019). An IoT Based Gradient Descent Approach for
Precision Crop Suggestion using MLP. Journal of Physics: Conference Series, 1362, 012038.
21. Faiz, M., & Daniel, A. K. (2022). A Multi-Criteria Dual Membership Cloud Selection Model based on Fuzzy
Logic for QoS. International Journal of Computing and Digital Systems, 12(1), 453-467.
22. Choudhary, S., Narayan, V., Faiz, M., & Pramanik, S. (2022). Fuzzy approach-based stable energy-efficient
AODV routing protocol in mobile ad hoc networks. In Software Defined Networking for Ad Hoc Networks
(pp. 125-139). Cham: Springer International Publishing.
23. Faiz, M., & Daniel, A. K. (2021, December). Multi-criteria based cloud service selection model using fuzzy
logic for QoS. In International Conference on Advanced Network Technologies and Intelligent Computing
(pp. 153-167). Cham: Springer International Publishing.
24. Narayan, V., Daniel, A. K., & Chaturvedi, P. (2023). E-FEERP: Enhanced fuzzy based energy efficient
routing protocol for wireless sensor network. Wireless Personal Communications.
25. Narayan, V., & Daniel, A. K. (2022). CHHP: coverage optimization and hole healing protocol using sleep
and wake-up concept for wireless sensor network. International Journal of System Assurance Engineering
and Management, 13(Suppl 1), 546-556.
26. Narayan, V., & Daniel, A. K. (2022). Energy Efficient Protocol for Lifetime Prediction of Wireless Sensor
Network using Multivariate Polynomial Regression Model.
27. Narayan, V., & Daniel, A. K. (2021, October). IOT based sensor monitoring system for smart complex and
shopping malls. In International conference on mobile networks and management (pp. 344-354). Cham:
Springer International Publishing.
28. Channi, Harpreet Kaur, et al. "Multi-Criteria Decision-Making Approach for Laptop Selection: A Case
Study." 2023 3rd Asian Conference on Innovation in Technology (ASIANCON). IEEE, 2023.
29. Faiz, Mohammad, et al. "Machine Learning Techniques in Wireless Sensor Networks: Algorithms,
Strategies, and Applications." International Journal of Intelligent Systems and Applications in
Engineering 11.9s (2023): 685-694.
30. Mall, Pawan Kumar, et al. "FuzzyNet-Based Modelling Smart Traffic System in Smart Cities Using Deep
Learning Models." Handbook of Research on Data-Driven Mathematical Modeling in Smart Cities. IGI
Global, 2023. 76-95
31. Narayan, Vipul, et al. "A theoretical analysis of simple retrieval engine." Computational Intelligence in the
Industry 4.0. CRC Press, 2024. 240-248.
 Akhil Panwar, et al / Kuey, 30(5), 4881 11041

32. Narayan, Vipul, et al. "A comparison between nonlinear mapping and high-resolution
image." Computational Intelligence in the Industry 4.0. CRC Press, 2024. 153-160.
33. Sandhu, Ramandeep, et al. "Enhancement in performance of cloud computing task scheduling using
optimization strategies." Cluster Computing (2024): 1-24.
34. kumar Mall, Pawan, et al. "Self-Attentive CNN+ BERT: An Approach for Analysis of Sentiment on Movie
Reviews Using Word Embedding." International Journal of Intelligent Systems and Applications in
Engineering 12.12s (2024): 612-623.
35. Narayan, Vipul, et al. "7 Extracting business methodology: using artificial intelligence-based
method." Semantic Intelligent Computing and Applications 16 (2023): 123.