Influence Analysis

Explore drivers of your project’s outcomes with influence analysis in Bayesian networks; identify influential nodes.

In this lesson, we conduct a thorough analysis to identify the drift factors that significantly influence the causal network. These factors are critical in steering the behavior and outcomes within our network and, hence, play a vital role in our analysis.

This initial analysis stage is crucial as it allows us to pinpoint the primary causes behind our specified project's problems. By identifying these influential drift factors, we can develop a deep understanding of the project's underlying issues and devise more targeted and effective solutions.

Influences on the final model

To evaluate the influence of one node over the target node "Overcost", we can perform ablation experiments. This involves removing the evidence for one node at a time and measuring the change in the marginal probability distribution of the "Overcost" node. The larger the change in the distribution, the more influential the node is. Here's how we can do this:

Get hands-on with 1300+ tech skills courses.