Assessment of Water Quality using Machine Learning and Fuzzy Techniques

The water quality of river Ganga is an important concern due to its drinking, domestic uses, irrigation and also for aquatic life. But the extent of pollutants in river water has deteriorated the quality of river water. So, the assessment of river water becomes very important. But due to the involved subjectivity and uncertainty in the decision making parameter makes the task very complex. In this study, machine learning and fuzzy techniques are utilized to develop the river water quality assessment models. The quality of the water is grouped into three classes. Four machine learning algorithms namely decision tree, random forest tree, k-nearest neighbor and support vector machine are used and implemented on python and anaconda platform. Whereas, three fuzzy based models (fuzzy decision tree, wang-mendel and fast prototyping) are developed using Guaje open source software. All the seven models are analyzed in terms of accuracy, precision, recall and f1-score. The observed result shows that the fuzzy decision tree-based assessment model performs more accurately as compared with the machine learning based models.