TY - JOUR
T1 - Unsupervised Fertigation and Machine Learning for Crop Vegetation Parameter Analysis
AU - Mohd Rahman, Mohd Izzat
AU - Mohd Razman, Mohd Azraai
AU - Abdul Majeed, Anwar P.P.
AU - Aiman Shapiee, Muhammad Nur
AU - Abdullah, Muhammad Amirul
AU - Musa, Rabiu Muazu
N1 - Publisher Copyright:
© 2023, Ismail Saritas. All rights reserved.
PY - 2023/7/16
Y1 - 2023/7/16
N2 - This study proposes an IoT-based smart irrigation management system that can optimize water-resource utilization in a smart agricultural system. The system uses unsupervised learning-based clustering to predict the irrigation needs of a field based on the ground parameters sensed by automated monitoring devices. These parameters include soil moisture, light intensity, temperature, and humidity. The system extracts feature such as the maximum, minimum, mean, and standard deviation of four soil moisture sensors from the primary dataset of plants. Then, it applies lag features to enhance the accuracy of the classification model. The system uploads the dataset of 108 features to the Orange GUI and performs k-means clustering to assign cluster labels to the data as meta-attributes in a new dataset. The study evaluates the system using a month’s worth of data and demonstrates its functionality and effectiveness. The system employs machine learning techniques such as Random Forest, Neural Network, and kNN, which achieve 100%, 99.9%, and 99.8% accuracy respectively.
AB - This study proposes an IoT-based smart irrigation management system that can optimize water-resource utilization in a smart agricultural system. The system uses unsupervised learning-based clustering to predict the irrigation needs of a field based on the ground parameters sensed by automated monitoring devices. These parameters include soil moisture, light intensity, temperature, and humidity. The system extracts feature such as the maximum, minimum, mean, and standard deviation of four soil moisture sensors from the primary dataset of plants. Then, it applies lag features to enhance the accuracy of the classification model. The system uploads the dataset of 108 features to the Orange GUI and performs k-means clustering to assign cluster labels to the data as meta-attributes in a new dataset. The study evaluates the system using a month’s worth of data and demonstrates its functionality and effectiveness. The system employs machine learning techniques such as Random Forest, Neural Network, and kNN, which achieve 100%, 99.9%, and 99.8% accuracy respectively.
KW - Chili Plant
KW - Classification
KW - Feature Extraction
KW - Fertigation System
KW - Machine Learning
UR - http://www.scopus.com/inward/record.url?scp=85174926479&partnerID=8YFLogxK
M3 - Article
AN - SCOPUS:85174926479
SN - 2147-6799
VL - 11
SP - 417
EP - 425
JO - International Journal of Intelligent Systems and Applications in Engineering
JF - International Journal of Intelligent Systems and Applications in Engineering
IS - 3
ER -