Information Panel

​Context

• The Information Panel is positioned at the top of the Monitor Panel.

• The Information Panel features up to eight different statistics.

• It is only visible in Validation Mode.

  

Usage

• You can open  and close  the Information Panel with the button located at the bottom right corner.

Content

Joint Probability

• The Joint Probability is calculated based on the current set of evidence, e.g., hard positive, hard negative,  virtual, soft, and numerical.
• The Joint Probability is also known as the Marginal Likelihood.
• Without any evidence set, the Joint Probability is 100%.
• BayesiaLab updates the Joint Probability automatically with each modification of evidence via the Monitors.

Log-Loss  

• Log-Loss is the binary logarithm of the Joint Probability.
• If the Joint Probability equals 100%, the Log-Loss is 0.
• The Log-Loss increases towards infinity as the Joint Probability approaches 0.
• Like the Joint Probability, the Log-Loss is automatically updated with each new piece of evidence.

Cases

• If a data set is associated with the network, the Information Panel shows an estimate of the number of records that correspond to the current set of evidence.
• If no evidence is set, the number reported by Cases is equal to the total number of records in the associated data set.
• Whenever evidence is modified via the Monitors, the value of Cases is automatically recomputed.

Total Value  

• The Total Value shown in the Information Panel represents the sum of the expected values of all nodes in the network given the current set of evidence.

Mean Value

• The Mean Value shown in the Information Panel represents the arithmetic mean of the expected values of all nodes in the network given the current set of evidence.

Expected Compression

• Expected Compression measures the relative reduction of the sum of the entropies of all unobserved nodes compared to the sum of their prior entropies.

Actual Compression

• Actual Compression measures the relative Information Gain of the current set of evidence, i.e., the Information Gain normalized by the Marginal Likelihood computed with the fully unconnected network.

Resources

• Resources shows the sum of all Utility Nodes.

Display Options

If not shown by default, you can add Compression and Resources to the Information Panel:

• Select Node Context Menu > Show > Compression.
• Select Node Context Menu > Show > Resources.

Workflow Animation