Prediction Model of Water Quality and Detection of Vibrio Cholerae Bacteria

This document presents the results of two tests related to water quality based on the physico-chemical characteristics provided by the dataset used;both tests were performed based on the same dataset from which the membership sets were defined,and the most relevant functions were defined. The objective of this work is to develop a prediction model for Water Quality based on the detection data of the Vibrio Cholerae Bacteria by comparing two methods, one focused on precision and the other on interpretability. The first test used a neural network to predict water quality using variables like pH, temperature, turbidity, and salinity; the second used a fuzzy logic system to detect Vibrio Cholerae in water using the usual variables associated with its presence: temperature, salinity, phosphate, and nitrite levels. There are two phases to this study's methodology. The first phase is the development of an adapted software using an iterative and incremental process model based on prototypes. The second phase or operational phase has an experimental characterization that allows an adaptation of the medium to establish the main characteristics and properties relevant to the object of study. The results showed efficacy values of 99.99% (highest value obtained) for the first trial and 70.23% for the second trail; these values represent an accurate prediction of water quality and valuable detection of cholera-related bacteria in water supplies. Through the graph of correspondences between the established rules and the membership functions in the input and output sets, this study has built two systems that are highly interpretable and transparent to people.