Bubble Chart Calculator for Machine Learning Graphs

Bubble Chart Results

Processed Data

#	Label	Category	Set	X	Y	Raw Size	Marker Size	Outlier	Error

Interpretation

Group Summary

Group	Count	Mean X	Mean Y	Total Size

Merged Results Data

Generate a chart to view results.

1. Data Input

Upload CSV

Expected columns: x, y, size, label, category, set, actual, predicted, misclassified, color, tooltip.

Upload JSON

Use an array of row objects.

Paste Spreadsheet or CSV Data

X	Y	Size	Label	Category	Dataset	Actual	Predicted	Error?	Color	Tooltip

2. Chart Labels and Appearance

Chart Title

Subtitle

X-Axis Label

Y-Axis Label

Bubble-Size Meaning

Chart Theme

Color Palette

Marker Symbol

Chart Height

Bubble Opacity: 0.75

Border Width: 1

Show legend

Show gridlines

Show bubble labels

Show detailed tooltips

Use row colors

Animate rendering

Show crosshair spikes

Live update controls

3. Bubble Size and Axis Controls

Bubble Size Source

Size Expression

Allowed: x, y, size, abs, sqrt, log, min, max, pow.

Scaling Method

Minimum Marker Size

Maximum Marker Size

Displayed Decimal Places

Number Format

Tooltip Prefix

Tooltip Suffix

X Minimum

X Maximum

Y Minimum

Y Maximum

Logarithmic X-axis

Logarithmic Y-axis

Clamp non-positive sizes

Reduce label overlap

4. Filtering, Grouping, and ML Analysis

Coordinate Mode

Category Filter

Dataset Split

Search Labels

Filter X Minimum

Filter X Maximum

Filter Y Minimum

Filter Y Maximum

Sort Observations

Observation Selection

Top or Bottom Count

Outlier Method

Outlier Threshold

Hover Mode

Linear regression trendline

Highlight outliers

Highlight error rows

Aggregate duplicate coordinates

Remove invalid rows

Calculate group summaries

Calculate weighted means

Actual versus predicted metrics

Formula Used

Each row supplies an X coordinate, Y coordinate, and magnitude. Marker sizes are mapped between selected minimum and maximum limits.

Linear scaling: m = m_min + ((s − s_min) ÷ (s_max − s_min)) × (m_max − m_min).

Correlation: r = Σ((x − x̄)(y − ȳ)) ÷ √(Σ(x − x̄)²Σ(y − ȳ)²). Regression uses ordinary least squares.

How to Use

Enter rows manually, paste data, or upload CSV or JSON.
Choose labels, scaling, axes, colors, filters, and analysis options.
Generate the chart and inspect statistics, groups, warnings, and selections.
Export the visualization or processed dataset in your preferred format.

Example Data Structure

x	y	size	label	category	set	actual	predicted
1.2	2.8	45	Model A	Cluster 1	Training	2.7	2.8
3.4	4.3	110	Model C	Cluster 2	Testing	4.8	4.3

Frequently Asked Questions

What does bubble size represent?

It represents a third numeric variable such as confidence, frequency, error, sample weight, or importance.

Should marker radius or area represent values?

Area-based scaling is often easier to interpret because viewers compare visible marker areas.

Can categories use different colors?

Yes. Category groups receive palette colors, or each row can provide a custom color.

How are outliers detected?

The calculator supports Z-score distance across coordinates or IQR detection using bubble-size values.

Can I compare actual and predicted values?

Yes. Add actual and predicted columns to calculate MAE, MSE, RMSE, and R-squared.

What happens to invalid rows?

They are removed when validation filtering is enabled. Otherwise, generation stops with an error.

Can duplicate coordinates be combined?

Yes. Duplicate X and Y positions can be aggregated while their bubble sizes are summed.

Does the chart support zooming and selection?

Yes. Plotly provides zoom, pan, box selection, lasso selection, reset, hover, and click inspection.

Which export formats are available?

The chart exports to PNG, SVG, or PDF. Processed data exports to CSV or JSON.