Relationship
Key influencers
- Relationship
What it shows
An automatic view of what most affects a measure.
When to use it
Use an explanatory model to identify variables associated with a target outcome when you have adequate observations and relevant predictors.
Worked example
For late shipments, a model might associate a particular carrier and route with higher delay rates. That does not prove the carrier caused the delay.
Data required
A clearly defined target, predictor fields available at the time of decision, sufficient examples and a documented data collection process.
Checks and common mistakes
Inspect class imbalance, missing values, leakage, correlated predictors and subgroup performance. Compare the result with a transparent baseline and domain evidence.
More accessible alternative
Use segmented rates and a simpler model when the advanced visual cannot be explained or audited.