When should I use a scatter plot?

Exploring correlation between two variables ● Spotting outliers and clusters ● Regression analysis

When should I avoid a scatter plot?

Categorical data ● Evenly-sampled time series (a line chart is usually clearer) ● Too many overlapping points (>1000) — use hexbin or 2D contour

What are common mistakes with scatter plots?

Assuming correlation = causation ● Overplotting without transparency ● Ignoring outliers

Relationship

Scatter Plot

Plots two numeric variables as points, revealing correlations, clusters, and outliers.

Study Hours vs Test Score

50 students, midterm exam

View data (50 rows)

Chart data table: Study Hours vs Test Score
Hours	Score
1	35
1	41
1	38
2	42
2	45
2	47
2	40
3	46
3	50
3	53
3	48
4	51
4	55
4	58
4	53
5	56
5	60
5	62
5	58
5	55
6	60
6	64
6	67
6	62
6	65
7	65
7	69
7	72
7	67
7	70
8	71
8	75
8	78
8	72
8	74
9	76
9	80
9	82
9	78
9	79
10	81
10	84
10	88
10	83
10	86
11	87
11	90
11	92
12	93
12	95

Make a scatter plot with your data

Use a scatter plot when…

Exploring correlation between two variables
Spotting outliers and clusters
Regression analysis

Avoid when…

Categorical data
Evenly-sampled time series (a line chart is usually clearer)
Too many overlapping points (>1000) — use hexbin or 2D contour

Data it needs

Property	Value
Min Rows	10
Min Columns	2
Column Types	numbernumber

Visual anatomy

Marks

circle

Channels

position-xposition-ycolor-hue (optional)size (optional, for bubble variants)

Axes

x-quantitativey-quantitative

Guiding principles

Consider instead

Bubble Chart

To add a third variable via size

Hexbin Plot

For dense data with overplotting

Common mistakes

Assuming correlation = causation
Overplotting without transparency
Ignoring outliers

History

Pioneered by John Herschel in the 1830s; formalized by Francis Galton in the 1880s.

Accessibility notes

Report the correlation coefficient and describe trend as positive/negative/none. When using color to encode a third variable, pick a colorblind-safe palette and pair color with shape or label so the encoding survives a grayscale print.

Got data? Let's see what works.

Drop your CSV. You'll get a Scatter Plot plus four alternatives - ranked by which one actually fits your data best.

Try it free

Hours	Score
1	35
1	41
1	38
2	42
2	45
2	47
2	40
3	46
3	50
3	53
3	48
4	51
4	55
4	58
4	53
5	56
5	60
5	62
5	58
5	55
6	60
6	64
6	67
6	62
6	65
7	65
7	69
7	72
7	67
7	70
8	71
8	75
8	78
8	72
8	74
9	76
9	80
9	82
9	78
9	79
10	81
10	84
10	88
10	83
10	86
11	87
11	90
11	92
12	93
12	95

Hours	Score
1	35
1	41
1	38
2	42
2	45
2	47
2	40
3	46
3	50
3	53
3	48
4	51
4	55
4	58
4	53
5	56
5	60
5	62
5	58
5	55
6	60
6	64
6	67
6	62
6	65
7	65
7	69
7	72
7	67
7	70
8	71
8	75
8	78
8	72
8	74
9	76
9	80
9	82
9	78
9	79
10	81
10	84
10	88
10	83
10	86
11	87
11	90
11	92
12	93
12	95