Opening the jar file should show you the Tetrad interface. Please see Tetrad for documentation on the interface. The new buttons we have added are dbData and dbEstimator.
- dbData allows you to select an input database for structure learning, and set the database connection parameters using the pull-down menu under "File".
- dbEstimator produces a Functor Bayes net with conditional probability parameters for the input database. Learning can be run as follows.
- Draw an arrow from "dbData" to "dbEstimator".
- Click on "dbEstimator".
- The database is expected to consist of entity and relationship tables (ER model). The primary key of the entities should should end with "_id".
- Foreign key constraints should be set for relationship tables. The foreign keys should cascade on delete.
To perform class-level Inference using Bayes nets.
From Tetrad's point of view, the output of dbEstimator is a full parametrized Bayes net (a Bayes IM) that can be connected to their inference routines (the "Updater").
There are various tools in the pull-down menus for converting the Bayes nets to Markov Logic networks. We explain this in the Markov Logic Network pages.
Examples.You should be able to load the files below into the Tetrad session so you can see the complete setup for a sample database.
- Mondial Tetrad File
- University Tetrad File. (Some problem with Updater).
- MovieLens Tetrad File
- Hepatitis Tetrad File
- Make graph layout changeable in the Updater.
- Change database column names away from "dummy".
- Learn on full Mondial database. (Make "Borders" symmetric.)