Scatter Plot Calculator

Scatter Plot Results

Relationship interpretation

Regression equation

Selected points

Use box or lasso selection.

Warnings and notes

Observation Table

1. Data Input

Enter data manually or upload a CSV file. The first row may contain headers.

Data rows

Columns can include X, Y, Group, Label, Size, Color, Predicted, Split, and Confidence.

CSV or TXT file

Delimiter

Header row

2. Column Mapping and Cleaning

X-axis column

Y-axis column

Group or class column

Point label column

Bubble size column

Continuous color column

Predicted value column

Train, validation, or test column

Confidence column

Missing value handling

Duplicate observations

Maximum rows

3. Plot and Point Options

Scatter plot type

Marker shape

Base marker size

10 pixels

Marker opacity

0.80

Marker border width

Jitter amount

Point labels

Highlight points

Bubble size scale

5. Trendline and Regression Options

Trendline type

Polynomial degree

Moving average window

LOESS span

0.50

Confidence level

Trendline extension

Percent beyond observed X range.

Display equation

Display R²

Show confidence interval

Show prediction interval

6. Transformations and Outlier Analysis

X transformation

Y transformation

Outlier method

Outlier threshold

Outlier action

Largest errors to highlight

7. Reference Lines and Regions

Vertical reference value

Horizontal reference value

Reference line label

Acceptable region X minimum

Acceptable region X maximum

Acceptable region Y minimum

Acceptable region Y maximum

8. Output and Export Preferences

Show statistical cards

Show observation table

Include invalid rows in table

Auto-download results after five seconds

Formula Used

Pearson correlation:

r = Σ[(xᵢ − x̄)(yᵢ − ȳ)] ÷ √(Σ(xᵢ − x̄)² × Σ(yᵢ − ȳ)²)

Linear regression:

ŷ = b₀ + b₁x

b₁ = Σ[(xᵢ − x̄)(yᵢ − ȳ)] ÷ Σ(xᵢ − x̄)²

b₀ = ȳ − b₁x̄

Coefficient of determination:

R² = 1 − Σ(yᵢ − ŷᵢ)² ÷ Σ(yᵢ − ȳ)²

How to Use

Paste data or upload a CSV file.
Select the X and Y columns.
Choose groups, labels, sizes, or predictions.
Configure trendlines, transformations, and outlier rules.
Customize axes and chart appearance.
Select Create Scatter Plot.
Inspect points, statistics, and warnings.
Export the graph, table, or report.

Example Data

X	Y	Group	Label	Size	Predicted	Split
1.0	1.4	A	Sample 1	8	1.3	Train
4.0	4.3	B	Sample 4	16	4.0	Train
7.0	6.7	C	Sample 7	14	7.2	Validation
10.0	10.2	C	Sample 10	26	9.8	Test

Understanding Scatter Plots

A scatter plot displays paired numeric observations. Each point represents one data record. Patterns reveal relationships between two variables.

Upward patterns suggest positive association. Downward patterns suggest negative association. Curved patterns may require nonlinear models.

Outliers can strongly affect correlation. Groups can hide different local patterns. Correlation alone never proves direct causation.

Machine learning teams use scatter plots often. They inspect predictions, residuals, clusters, and feature relationships. Clear labels support reliable model diagnosis.

Frequently Asked Questions

What does a scatter plot show?

It shows relationships between two numeric variables. Every marker represents one observation. Visible patterns support exploratory analysis.

What does a positive correlation mean?

Both variables generally increase together. Stronger patterns produce larger positive coefficients. Causation still requires separate evidence.

What does a negative correlation mean?

One variable generally decreases as another increases. The coefficient approaches negative one. Outliers may change the result.

When should I use Spearman correlation?

Use Spearman for monotonic relationships. It works with ranks instead. It handles nonlinear ordering better.

Why can R² be misleading?

R² measures explained variation only. It cannot confirm model correctness. Residual inspection remains important.

How are outliers detected?

The calculator supports several methods. IQR and Z-scores inspect distributions. Residual methods inspect regression influence.

Can I compare machine learning splits?

Yes, map a split column. Train, validation, and test points separate. This exposes distribution differences.

Can I create a bubble chart?

Yes, map a size column. Bubble area represents another variable. Keep size values positive.

Does correlation prove causation?

No, correlation only measures association. Confounders may create apparent relationships. Experimental evidence supports causal conclusions.