.. index:: Prediction

.. _Prediction:

Prediction
==========

In a :ref:`DBN<Dynamic_Bayesian_Network>` we can compute beliefs for nodes in time slices that lie beyond the time window. This operation is called prediction.

.. Note:: only DBNs entirely consisting of discrete nodes supports prediction.

To perform prediction, click the |figure1| prediction button and summon the prediction wizard.

.. |figure1| image:: images/predictiontool.png

Prediction Wizard
-----------------

The belief prediction wizard has two tabs for configuring the computations. First configure the number of time slices beyond the current time window to predict.

.. figure:: images/prediction.png

	Figure 1: Prediction, configure number of time slices beyond window to predict.

Next pick the nodes we wish to predict.

.. figure:: images/prediction1.png

	Figure 2: Prediction, pick which nodes to predict.

Finally when done configuring click the OK button to compute predictions. When computation is done a :ref:`data frame<data_frame>` appears with the predicted beliefs.

.. figure:: images/prediction2.png

	Figure 3: A data frame displaying the predicted beliefs. Each row represents a time slice. The first 0 to (n - number of slices to predict) rows are the time slices from the current window and the remaining rows are the predicted time slices.