Child pages
  • Arc Confidence (7.0)

Contents

Context

Tools | Resampling | Arc Confidence

Arc Confidence measures the confidence/robustness/variability of the machine learned arcs.

One of the three resampling methods is used for generating  data sets  that are thus utilized for learning  networks .

The frequencies of the arcs are computed and the structures are compared with the current network .

If the associated data set is split into Learning and Test sets, and respectively, the observations contained in are excluded from this analysis!

New Feature: Bootstrap

As of version 7.0, Bootstrap is now available for generating the data sets. 

Example

Here is the 2D-Mapping that highlights the strength of the relationships.

Below is the Relationship Analysis Report that allows getting the p-values of these relationships.

Given that we have 94 observations, Jackknife cannot generate more than 94 data sets. We chose for this method and for Bootstrap and Data Perturbation.

For each method, we extract the network with the following criteria:

We also display:

  • The Synthesis Structure. It graphically summarizes all the structures that have been induced. The thickness of the link is proportional to its frequency. Furthermore, the links in blue are those that are not in the current network ,
  • An extract of the Analysis Report to get the frequencies of the arcs of .

Extracted Network

Synthesis Graph

Analysis Reports

Extracted Network

Synthesis Graph

Analysis Reports

Extracted Network

Synthesis Graph

Analysis Reports