Scatter Diagram & Correlation
Plot paired observations, estimate the strength and direction of association and reveal clusters, curvature or outliers that deserve investigation.
Purpose and method
Plot paired measurements to explore correlation, direction, strength and possible outliers.
When to use it
Use a scatter diagram when a suspected input factor may influence a measurable output.
Core principle
Correlation describes association, not causation. A credible conclusion requires correct pairing, sufficient range and control of confounding factors.
What the interactive module provides
- XY chart
- Correlation coefficient
- Regression line
- scatter plot, correlation estimate and visible clusters or outliers
Professional verification: The module supports learning, analysis and documentation. Verify results against actual process conditions, internal standards, safety requirements and competent professional judgement before implementation.
01
Recommended workflow
Plot paired observations, estimate the strength and direction of association and reveal clusters, curvature or outliers that deserve investigation.
- 1
Define a plausible relationship and pairing rule
- 2
Collect X and Y at the same observation level
- 3
Plot all points before calculating a coefficient
- 4
Check range, clusters, curvature and outliers
- 5
Run a controlled test before claiming causation
02
Practical example
Weld current and nugget diameter appear correlated, but colour-coding by electrode age shows two separate populations and identifies the hidden factor.
Calculation or verification principle
Pearson r measures linear association from −1 to +1. Always inspect the plot because strong non-linear patterns or outliers can make r misleading.
Required inputs → Expected outputs
- Required inputs
- paired numeric X and Y observations collected from the same unit or event, plus optional grouping variable
- Expected outputs
- scatter plot, correlation estimate and visible clusters or outliers
03
Common mistakes to avoid
×
Pairing values from different times or parts
×
Using a narrow range that hides the relationship
×
Removing outliers without investigating them
×
Assuming a high r proves a causal mechanism
04
Frequently asked questions
What data do I need to use this tool?
Prepare paired numeric X and Y observations collected from the same unit or event, plus optional grouping variable.
What result does the module produce?
The module produces scatter plot, correlation estimate and visible clusters or outliers.
Can I implement the result without further review?
No. Treat the output as a structured engineering and improvement aid. Verify source data, assumptions, safety, quality, legal requirements and local process conditions with competent responsible people before implementation.
LEAN Toolbox
Open interactive tool →Scatter Diagram & Correlation
Plot paired observations, estimate the strength and direction of association and reveal clusters, curvature or outliers that deserve investigation.