Rug Plot Calculator

Rug Plot Result

Waiting for data

Enter values or load sample data, then select Generate Rug Plot.

1. Data Input

Numeric data

Use one value per line, or value and group separated by comma, tab, semicolon, or pipe.

Upload CSV or TXT Delimiter

First uploaded row contains headings

2. Plot Configuration

Orientation

Rug position

Mark style

Plot title

Primary axis title

Legend title

Chart width

Chart height

Plot theme

Primary axis scale

Custom axis minimum

Custom axis maximum

Approximate tick count

Custom reference value

Reference label

3. Rug Mark Controls

Mark height or size

Current: 15

Line thickness

Current: 2

Mark opacity

Current: 0.8

Jitter amount

Current: 0.15

Group offset

Current: 0.3

Duplicate handling

Primary mark color

Group palette

Line style

4. Overlays and Distribution Options

Histogram overlay

Kernel density curve

Box plot overlay

Violin plot overlay

ECDF curve

Histogram bin method Custom bins

Density bandwidth Custom bandwidth

Kernel

Histogram scale

Overlay opacity

Current: 0.45

5. Data Processing

Transformation

Outlier detection

Outlier threshold

IQR multiplier, z-score limit, or modified z-score limit.

Custom lower limit

Custom upper limit

Decimal places

Minimum filter

Maximum filter

Observation order

Exclude outliers from plot

Highlight outliers

Remove duplicates

Mean reference line

Median reference line

Quartile reference lines

Show gridlines

Show legend

Reverse primary axis

6. Generate and Manage

Detailed Results

Index	Original	Transformed	Group	Frequency	Percentile	Outlier	Duplicate	Included

Formula Used

Each rug mark is positioned at an observed numeric value. The mark location follows the selected axis and transformation.

Rug position = T(xᵢ)

Gaussian KDE:
f̂(x) = 1 / (n × h) × Σ K((x - xᵢ) / h)

IQR outlier limits:
Lower = Q1 - k × IQR
Upper = Q3 + k × IQR

The density estimate uses the selected kernel and bandwidth. Outlier rules can use IQR, z-scores, modified z-scores, or custom limits.

How to Use

Paste numeric values or upload a CSV or TXT file.
Select value and optional group columns after uploading data.
Choose rug orientation, mark style, overlays, and colors.
Apply transformations, filters, and an outlier detection method.
Select Generate Rug Plot to calculate and visualize results.
Review summary statistics, distribution findings, and observation details.
Copy results or export CSV, PNG, SVG, and printable PDF reports.

Example Data

Value	Group	Interpretation
12.5	Train	Lower-range training observation
18.2	Train	Central training observation
22.7	Test	Central testing observation
41.8	Test	Possible high-value outlier

Machine Learning Applications

Rug plots reveal feature coverage, repeated values, sparse regions, and distribution gaps. They support train-test comparison, class imbalance inspection, residual analysis, and distribution-shift checks.

Grouped marks can compare classes, clusters, predictions, and validation splits. Density, histogram, box, violin, and ECDF overlays add complementary context.

Frequently Asked Questions

What is a rug plot?

A rug plot places a short mark for every observation along an axis. Dense mark clusters indicate concentrated values.

Why use rug plots in machine learning?

They reveal feature coverage, duplicates, outliers, sparse areas, and class distribution differences before or after model training.

Can this calculator compare groups?

Yes. Add group labels manually or select a CSV group column to create colored group traces and summaries.

How are duplicate values handled?

You can keep, stack, merge, remove, or scale duplicate marks according to their observed frequency.

Which bandwidth should I choose?

Silverman works well for many datasets. Scott is another automatic choice, while custom bandwidth offers direct smoothing control.

How are outliers detected?

The calculator supports IQR limits, standard z-scores, modified z-scores, and user-defined lower and upper limits.

Can I transform values before plotting?

Yes. Available transformations include standardization, normalization, logarithms, square roots, centering, percentages, and winsorization.

What happens with invalid rows?

Nonnumeric or incomplete rows are rejected. The result message reports valid, filtered, excluded, and rejected observations.

Can I export the chart?

Yes. Export the plot as PNG or SVG, download results as CSV, and print the report to PDF.

Are large datasets supported?

Yes, but dense marks may overlap. Use merging, frequency scaling, lower opacity, or filtering for clearer interpretation.