Configure the Plot
Paste aligned observations, then customize visual and statistical options.
Formula Used
The calculator preserves every numeric value while shifting its displayed category position.
Displayed position = category index + (seeded random value × jitter range)
Category mean = Σxᵢ / n
Sample deviation = √[Σ(xᵢ − x̄)² / (n − 1)]
Category z-score = (xᵢ − category mean) / category deviation
A point becomes flagged when its absolute z-score exceeds your threshold.
How to Use This Calculator
- Paste one category label for every numeric observation.
- Paste matching values using any supported separator.
- Add optional series names and individual point labels.
- Choose sorting, orientation, jitter, summaries, and outlier settings.
- Submit the form and inspect results above it.
- Download the chart or category summary when finished.
Frequently Asked Questions
1. What does jitter change?
Jitter changes only the displayed category coordinate. It never changes numeric measurements, category membership, means, medians, or calculated deviations.
2. Which separators can I use?
You can separate items with commas, new lines, semicolons, tabs, or pipes. Empty items are removed before validation.
3. Why must input counts match?
Every value needs one category label. Optional series and point labels may contain one shared item or matching observation counts.
4. How are potential outliers identified?
The calculator computes z-scores within each category. Points exceeding the selected absolute threshold receive larger markers with dark outlines.
5. What do interquartile boxes show?
Each box spans the first and third quartiles. Its internal line marks the category median, helping reveal central spread.
6. When should means be connected?
Connect means when categories possess a meaningful sequence. Avoid lines for unrelated nominal groups because connections may suggest false continuity.
7. Can I compare multiple model series?
Yes. Provide one series label per observation, and the calculator creates separate traces with distinct colors and legend entries.
8. Why use a fixed jitter seed?
A fixed seed reproduces identical point offsets after repeated submissions. This consistency helps compare styling changes without shifting observations.
9. Can I export the results?
Use the chart toolbar for image exports. The summary button downloads category statistics as a CSV file for further analysis.
Understanding Categorical Scatter Plots
Core Purpose
Categorical scatter plots compare numeric observations across named groups. Each point preserves an individual measurement within its category. Jitter separates overlapping points without changing their measured values.
This chart reveals spread, clustering, gaps, and unusual observations. Group summaries add context without hiding the underlying samples. Analysts can compare categories before selecting suitable models or features.
Position and Summary Calculations
The category coordinate begins with an assigned integer position. Random jitter adds a small horizontal or vertical offset. The displayed coordinate follows position plus jitter multiplied by range.
Category means divide each group total by its observation count. Medians use the central ordered value for robust comparison. Sample deviation measures typical distance from each category mean.
Outlier detection uses standardized distance from the relevant category mean. The calculator divides each residual by category standard deviation. Points exceeding the chosen threshold receive a distinct outline.
Preparing Inputs
Enter categories and values using matching item counts. Optional series labels create separate traces and legend entries. Point labels can identify samples during hovering or direct display.
Choose orientation, sorting, jitter, size, opacity, and symbols. Summary options can display means, medians, or connected category averages. Axis limits support focused inspection of important value ranges.
Interpreting Results
Review the overview cards before interpreting the plotted distribution. Compare category counts because uneven samples may distort impressions. Inspect medians when extreme values strongly influence category means.
Use series grouping to compare models, cohorts, or experiments. Connected means can reveal directional changes across ordered categories. Avoid connecting categories when their order has no meaning.
Jitter improves visibility, but excessive jitter may blur category membership. Smaller markers help dense datasets remain readable and precise. Higher opacity emphasizes points but increases overlap in crowded groups.
Data Quality and Accessibility
Clean category names before plotting to prevent duplicates. Trim whitespace and standardize capitalization across records. Missing values deserve review instead of silent conversion.
Balanced categories make comparisons easier and more trustworthy. Large groups naturally appear denser than smaller groups. Count summaries prevent density becoming mistaken for stronger variation.
Color palettes remain readable on white backgrounds. Distinct symbols improve access when colors appear similar. Legends become especially useful when several series overlap closely.
Machine Learning Applications
Categorical scatter plots support exploratory machine learning analysis. They expose feature behavior across classes, segments, and outcomes. These patterns can guide transformations, encoding choices, and validation.
Export the chart after verifying labels, scales, and summaries. Downloaded images support reports, presentations, and model documentation. Clear settings produce reliable comparisons for practical decisions today.