Child pages
  • Evidence Data Weighting (5.3)

Contents

Context

Validation Mode | Data | Save Data | Generate Weights According to Evidence

New Feature: Weight Generation Based on Evidence Set

This function uses any evidence set on the current network to compute an observation weight for each row of the data set.

If you subsequently load the data set generated this way, you will obtain the exact marginal distributions that you previously defined by setting evidence.

Example

Let's take the VisitAsia network with the following pieces of Hard Evidence and Soft Evidence, which are set on a subset of 4 nodes:

Right-clicking on the Database Icon or using Data | Save Data brings up the Saving Database options windows:

When loading the previously-saved file, we define the Weight column as Weight.

Thus, we obtain the following distributions for the 10,000 cases in the new, weighted dataset.