Bank Customer Churn — CPT Explorer

Query Explorer Pathway Decomposer Markov Blanket CPT Explorer

Start here

Open the recommended scenario for this case
Adjust observed evidence or intervention settings
Move to a second tool without losing context
Compare obs() versus do() where available
Inspect paths, blankets, or CPT structure to explain the shift

Tools

Why CPT inspection matters

The conditional probability table is where the model meets the world. Every causal claim a Bayesian network makes — every intervention prediction, every counterfactual estimate, every diagnostic posterior — flows through these numbers. Inspecting them directly is the only way to verify that what the model encodes matches what the domain experts actually believe.

A row in the CPT specifies P(child | parent configuration). When the row reads counter to expert intuition, one of three things is true: the structure is wrong, the elicitation was incomplete, or the expert intuition itself was mistaken. Each of these has different remedies — and each is worth catching before the model is used to support a decision. The explorer below lays every row out in the same format so the rows that need attention become visually obvious.

Model notes

Node