Relationship
Scatterplot
Positions observations by two numeric variables to reveal relationships, clusters and outliers.
- Plot
- Relationship
- Outlier detection
What it answers
Are two numeric variables related?
- Shape
- Unconnected XY points
- What it shows
- Association, clusters, gaps and outliers.
Example
The question
Do cloudier evenings bring more ticket refunds?
Cloud cover and refunds, 12 evenings (illustrative)
Illustrative teaching data. It belongs to no company.
The misleading version
Same question, drawn badly
Dots joined into a line
What goes wrong. Joining the dots suggests one evening led to the next. These are separate evenings. Leave scatter points unjoined.
When to use it
Use it when
Exploring paired numeric measurements.
Avoid it when
When categories need comparison or a causal claim lacks evidence.
How to read it
Read each observation's X and Y positions and inspect the cloud.
The data you need
- Layout
- Long table
- Fields
- X: number; Y: number
- One row is
- One observation with paired numeric measurements. What is the grain?
- Rules
- X and Y must refer to the same observation. Do not independently sort either column.
- Optional
- Optional ID,Group,Weight. Binning and smoothing settings belong to chart configuration.
- Preparation
- Align measurements by observation ID, select relevant variables and handle missingness. Scale or standardise only where the analytical method requires it.
- Units
- State the unit of every coordinate/variable. A unitless transformed or standardised scale must be identified; no common denominator is assumed.
Worked example
Points(2,10),(4,100),(6,50). The middle X has the highest Y; the three observations do not follow a simple increasing line.
| X | Y |
|---|---|
| 2 | 10 |
| 4 | 100 |
| 6 | 50 |