Web-Based Management System of water pollution using Classification Techniques

This paper describes an implementation of a web-based system using classification techniques for prediction of water pollution type and appropriate treatments depending on the water quality Index. The benefits of data mining lie in the extraction of new knowledge automatically from the raw data to progress decision making. (C4.5) The decision tree was used for classifying water quality into five classes using fourteen parameters according to the World Health Organization's requirements. These parameters are taken for each sample of water in each ten water stations that selected for the investigations. First two classes were suitable for drinking water while other classes were not, therefore two types of classification techniques ((c4.5) decision trees and artificial neural network, millstone machine learning technique) were suggested to produce a decision concerning the type of pollution and devise proposition for the treatment of pollution. The experiment was carried on a real database validated by (Iraqi Ministry of Environment) gathered from ten authenticated treatment stations. The results show that using C4.5 decision tree classifier found to be the better in terms of the execution time while using NNT algorithm gave slightly better results in terms of the accuracy and error percentages. Also, the work shows that the techniques of data mining have the prospect to fast predict of the water quality class, as long as the given data are a true representation of the scope of knowledge.