MONITORING AND ANALYSING WATER QUALITY USING WIRELESS SYSTEM AND MACHINE LEARNING
LAVANYA SINGH
, SHAIK ABDUL KALAM , DR SIVAKUMAR S
Random Forest, ANN, XGBOOST KNN, ADABOOST, Decision Tree and SVM
Water is the most essential need for all forms of life. Predicting water quality is of prime importance while formulating the environmental control plan and can contribute to better water resource conservation. Accurate projections of water quality are the testimony that can assist authorities in making prudent choices before predicament. The goal of this study was to provide a novel Machine Learning based model for water quality prediction and it aimed for comparative analysis of different Machine Learning algorithms on the available dataset while evaluating their accuracy. Intensive and comprehensive approach was performed to study the potability of water. A software simulation using three sensors namely temperature, turbidity and pH sensor on Proteus platform was implemented initially. Then in the second stage of the study, a hardware project using these 3 sensors was set up with Arduino and ESP8266 to test real-time various water implies and store these values in cloud server using Node MCU and predict their potability. An extensive study was carried out to explore further the categorization of potable water using various Machine Learning algorithms in the third and final stage of the current project. Balancing the data set using re-sampling and shuffling helped to improve the accuracy of the models and prevent bias towards the majority class. The Random Forest algorithm (RF) was executed the best with an accuracy of 88 percent. The Decision Tree and XGBOOST algorithms also performed well, achieving accuracies of 80 percent and 86 percent respectively. The SVM and ANN algorithms performed inferiorly, achieving accuracies of 70 percent and 68 percent respectively. The KNN and AdaBoost algorithm under performed with 66 percent and 63 percent accuracies respectively Performance of the ML techniques were also evaluated using accuracy precision, recall, F1 Score and MCC score which reconfirmed the highest performance of RF algorithm. Performance of all the ML algorithms were compared against Deep Learning (ANN), it was found all the tree-based classifiers (ML algorithms) outperformed the Deep Learning algorithm (ANN). with RF showing the best accuracy of 88 percent. It can also be reasonably concluded and deduced that deep learning algorithms have limited performance on tabular and numerical data which is not linearly separable. DL algorithms are superior for images and text. These results suggest that the RF algorithm isa promising approach for classifying potable water and can be used for future research in this area. Asa future scope the IOT based hardware system of this present study can be explored to testing real-time water samples and analysing their potability.
"MONITORING AND ANALYSING WATER QUALITY USING WIRELESS SYSTEM AND MACHINE LEARNING", IJSDR - International Journal of Scientific Development and Research (www.IJSDR.org), ISSN:2455-2631, Vol.8, Issue 10, page no.617 - 628, October-2023, Available :https://ijsdr.org/papers/IJSDR2310102.pdf
Volume 8
Issue 10,
October-2023
Pages : 617 - 628
Paper Reg. ID: IJSDR_205773
Published Paper Id: IJSDR2310102
Downloads: 000347268
Research Area: Engineering
Country: -, -, --
ISSN: 2455-2631 | IMPACT FACTOR: 9.15 Calculated By Google Scholar | ESTD YEAR: 2016
An International Scholarly Open Access Journal, Peer-Reviewed, Refereed Journal Impact Factor 9.15 Calculate by Google Scholar and Semantic Scholar | AI-Powered Research Tool, Multidisciplinary, Monthly, Multilanguage Journal Indexing in All Major Database & Metadata, Citation Generator
Publisher: IJSDR(IJ Publication) Janvi Wave
