Water pipes leak prediction in QWAT databases

Romanian water supplying companies are involved in a daily activity that consumes resources, both hardware and man power. That activity(ies), is finding hardware failures (mostly leaks) and repairing them. Such activities are not specific to Romanian companies but to all water supplying companies in all the countries. Researches in this field had been done before and highlights that companies are better to prevent leaks than to stop them, but to prevent leaks they need a way of predicting them prior their occurrence. This study wants to present a way of predicting leaks in open source databases specific for water supplying companies. The article focuses on preparing the QWAT (Acronym from Quantum GIS Water Plugin) data to be neural network friendly and how to use that data to make predictions. Other purpose of this article is to show that a web-based prediction tool can be built using open source resources.