Alice

Alice is a Data Mining tool produced by Isoft (www.isoft.fr). Alice can help the business user to find hidden trends in his data, and to extract new models, validated by his own knowledge, from these trends. And with its client/server architecture, ALICE/Server can now mine heaviest databases.

To build decision tree from data by Alice, several steps should be followed. Firstly, data source should be figured out. User can extract data from a lot of data sources with different format, such as Text file, Oracle Database, Microsoft Excel, and so on, and the fields of the dataset can be assigned if the data are come from plain text. Secondly, some preprocessing can be done on the collected data, you can create calculate fields, group values of a symbolic field, and also you can discretize numeric field. After the preprocessing, decision tree can be built based on these cleansed data, and the result can be displayed in graphic. Once the decision tree has been constructed, advanced analysis can be done, including correlation, crosstab, stat, and report. User can find more knowledge by doing such analysis on the decision tree. The decision tree can be tested and used in Alice, and the system will print out the test results into report for future refinement.

Alice provides user-friendly interface, so it is very easy to familiar with the operation, and it doesn’t require user have advanced computing and statistics background. But Alice can’t build decision tree with multi-variables, and it doesn’t allow user to designate the pruning method, or maybe it does no pruning during tree building. In Alice, the decision tree can’t be output by way of rule, and if it does so, the output maybe more understandable to user.

In short, Alice is a Interactive Exploratory System to find Classification from Data by using of Machine Learning(Decision Tree). Although there are still some shortcomings in the system, it is a very good data mining tool.