Modelling Uranium in Vicinity of Groundwater Population by Neural Networks of Multilayers Perceptron

The existent of Uranium in vicinity of groundwater population can give a threat to the water supplier for human consumption. The objective of the research was to find the most important variables to the existence of the Uranium. This paper shows some modelling process for above matters by applying Neural Networks of Multilayers Perceptron. Data taken from US Department of Energy. Neural Networks used in this study were learning the representation of the model inside the data, and how best it relation with the output variable that we obtained from prediction. The results showed that the training samples was 87 out of 127, and the testing samples was 40 out of 127. The results were not giving indication that a mathematical model obtained. The conclusion was Conductivity becoming the most important variable to the existence of Uranium, which followed by the second importance that was Arsenic, the third importance was Selenium, the fourth important was Total Alkalinity.