Decision support systems are being used in many areas and their use is still increasing. But one of their problems is that most decision support systems can only be used on computers, so they are sometimes not available where the decision needs to be made. For example this can happen where decision support systems are used to set medical diagnoses. Ten years ago this problem was solved by Blaž Zupan, Aleš Porenta and Gaj Vidmar with their Decisions at Hand system, which worked on Palm OS PDAs of the time. But since then Palm OS PDAs have practically disappeared and were replaced by smartphones. The goal of this paper is to implement and present a system similar to Decisions at Hand on the Symbian mobile operating system. The application is a decision support shell which means the machine learning takes place separately on a computer, where the computed model is written into an XML file. On the phone the XML file is used to build a user interface which supports the input of variables and the display of results. The phone application supports variable transformations and calculations. It also supports two decision making methods: the naive Bayes classifier and logistic regression.
|
