A method for managing oilfield water. Oilfield water data is grouped into discrete and non-overlapping groups. Outliers are removed from the data. Features of the remaining data are analyzed to identify most discriminative feature. The data is separated into training data and testing data, and the training data is fit into a model that shows the best precision, accuracy and recall. The model is confirmed using the testing data. Upon confirmation of the accuracy of the model, the model is applied to data for a new proposed oilfield well, and a new proposed project is implemented or disapproved based on a result of the identified model that predicts water production of the new proposed oilfield well.