An intelligent recommender system for drinking water quality

This article presents a recommender system based on rough mereology for the prediction of water quality in Louisiana, a state located in the southern region of the United States. The proposed system firstly maps the water dataset into a normalized dataset of potential of hydrogen (PH) levels prediction. Then, rough mereology and rough inclusion techniques are applied for clustering and classifying the normalized PH dataset into sets of granules with different radius. Voting by objects approach is subsequently applied in order to select the optimized granules. Finally, normalized rating matrix is acquired, then the predicted PH level will be recommended. The data, tested by the proposed recommender system, was collected from stations of the United states Environmental Protection Agency (EPA). The obtained results demonstrate the effectiveness and the reliability of the proposed recommender system. Based on the data resulted, the average PH level prediction in a certain time is characterized by a mean absolute error of 0.34. In addition, both experimentally resulted and actual dataset values existed in the healthy region of the PH level for drinking water, which is within the range 6.5 to 8.0 according to the World Health Organization (WHO) drinking water guidelines.