Frequency Table Maker: Counts, Percentages & Grouped Data
Turn numbers or categories into a frequency table with relative frequencies and percentages. Keep individual values or group numbers into intervals, then inspect the chart and export your results.
← All free toolsYour frequency table
Loading the numerical example…
Relative frequencies show 6 decimal places; percentages show 2. Rounded rows can total 99.99% or 100.01%. The total row uses unrounded counts.
How this table was calculated
Frequency Table Maker: Create and Understand Frequency Distributions
Use the tool above to count numerical values or categories. This guide explains relative frequencies, ordered cumulative totals and grouped intervals, with examples you can load into the calculator.
Key takeaways
The number of included observations in a value, category or interval.
The row count divided by all included observations; multiply by 100 for a percentage.
A running total that requires a meaningful numerical or categorical order.
Each value belongs to one interval. The final interval includes its upper boundary.
What Is a Frequency Table?
A frequency table counts how often each value or category appears. It keeps repeated observations together, so you can see common responses, rare values and gaps without scanning the raw list. An ungrouped table keeps each distinct value; a grouped frequency table combines numerical observations into intervals.
How to make a frequency table from your data
- Choose the interpretation. Use numerical values for measurements and counts. Use text categories for names, survey responses and identifiers such as “001.”
- Paste your observations. One value per line works well for a spreadsheet column. For several columns, choose the tab or CSV format, preview the data and select one column.
- Choose how to count. Keep exact values, or select grouped intervals for numerical data. Review any missing or invalid entries before excluding them.
- Generate and inspect. Check the included observation count, the table and its matching chart. Copy the table or download a CSV when the settings match your question.
CSV headings are optional: tick the header box only when the first row names the columns. Otherwise, the first observation stays in the dataset. If spellings or missing cells need attention, the CSV Data Cleaner can help you prepare a consistent source file.
Frequency, relative frequency and percentage
Frequency is a count. Relative frequency divides that count by the number of included observations: relative frequency = f ÷ n. Multiply the result by 100 to express it as a percentage. In an eight-observation dataset, a value appearing three times has relative frequency 3 ÷ 8 = 0.375, or 37.5%.
Percentage = 100 × f ÷ n
Here, f is the row count and n is the total number of included observations.
The denominator matters. If you exclude two missing observations from ten records, the percentages describe the remaining eight. They do not describe ten complete responses. The summary reports records, missing entries, invalid entries and exclusions so you can explain that choice when sharing results.
When cumulative frequency makes sense
Cumulative frequency adds the counts through the current row. For numerical values sorted from low to high, it tells you how many observations are at or below that value. Cumulative relative frequency divides the running count by n; the last row reaches 1, or 100%.
Categories such as Apple, Banana and Orange have no inherent low-to-high order. Their default table therefore has no cumulative columns. For ordinal responses such as Low, Medium and High, select your own category order and confirm that it is meaningful. Sorting nominal categories by frequency alone does not create an ordinal scale.
Grouped tables: class width, boundaries and midpoints
Exact-value tables are useful when a variable takes a few distinct values, such as the number of calls received. Measurements with many distinct values often need grouping. Grouping makes the overall distribution easier to inspect, but loses detail: a count in the interval from 10 to 20 does not tell you each original value.
This frequency distribution calculator offers automatic intervals, a chosen interval count, a class width with a starting boundary, and custom boundaries. Automatic grouping uses k = ceil(1 + log₂(n)), known as Sturges’ rule. It is a starting point. Try another interval count when a few bins conceal detail or many bins leave mostly empty groups.
Sturges’ rule suggests an interval count. You can choose a different count, width or set of boundaries.
Which interval contains a boundary value?
Each ordinary interval includes its lower boundary and excludes its upper boundary. With boundaries 0, 10, 20 and 30, a value of 10 enters the second interval. The final interval includes 30. Labels use inequalities to make this visible: 0 ≤ x < 10, 10 ≤ x < 20, and 20 ≤ x ≤ 30.
Class width is the upper boundary minus the lower boundary. The midpoint lies halfway between them. A midpoint describes an interval, not an observed measurement. Custom boundaries must increase strictly and cover every included value; the tool reports out-of-range values instead of dropping them.
Bar charts and histograms
Categories and exact values use separated bars. Exact-value bars are equally spaced labels, not a continuous numerical axis. Grouped data use a histogram with widths proportional to the intervals. This tool uses frequency density for every grouped chart: count divided by interval width. Bar area then represents frequency, including when widths differ. Read histogram versus bar graph for the distinction, or use the Histogram Maker for a chart-focused workflow.
Worked frequency table examples
Eight numerical observations
For 1, 2, 2, 3, 3, 3, 4, 4, the value 3 is the most frequent. Six of the eight observations are at or below 3. Load “Numbers” above to reproduce this table.
| Value | Frequency | Relative | Percent | Cumulative |
|---|---|---|---|---|
| 1 | 1 | 0.125 | 12.5% | 1 |
| 2 | 2 | 0.250 | 25% | 3 |
| 3 | 3 | 0.375 | 37.5% | 6 |
| 4 | 2 | 0.250 | 25% | 8 |
Fruit categories and unequal intervals
The fruit preset contains Apple three times, Banana twice and Orange once. Their percentages are 50%, 33.33% and 16.67%. Case is preserved unless you explicitly combine case variants; “Apple” and “apple” can remain separate.
The unequal-interval preset uses 1, 2, 8, 9, 10, 12, 18 and 25 with boundaries 0, 10, 20 and 40. Counts are 4, 3 and 1. Widths are 10, 10 and 20, so densities are 0.4, 0.3 and 0.05. The final bar is twice as wide, but its area represents only one observation.
Common Frequency Table Mistakes to Avoid
- Using the wrong denominator: Divide by included observations, not the number of distinct categories.
- Counting a boundary twice: An interior boundary belongs to the interval starting there.
- Adding unordered categories cumulatively: Nominal categories need counts and shares, not an assumed low-to-high order.
- Using count heights for unequal intervals: Use density so histogram area represents frequency.
- Entering thousands separators: In a comma-separated list, 1,200 is interpreted as two observations.
Methodology and limitations
The parser handles quoted CSV fields, embedded commas, escaped quotes and line breaks. Multi-column input requires a selected column. Empty numerical entries are never converted to zero. Blank interior lines and missing CSV cells are counted; trailing newlines in one-per-line mode are ignored. Missing categories use a separate internal identity, even if an actual label says “Missing” or “[Missing entry].”
Numerical counting uses browser double precision without rounding before grouping. Equivalent inputs such as 1, 1.0 and 1e0 count together; negative zero displays as zero. Values beyond safe integer precision and nonfinite numbers are rejected. Decimal class-width boundaries are built without repeated floating-point addition. Very close inputs can still share a browser representation; use specialist precision software when that distinction matters.
Limits are 1 MiB of text, 20,000 observations, 100 source columns, 2,000 distinct output values or categories, and 200 intervals. Tables show 50 rows per page. Charts show the first 30 rows in the selected order for ungrouped data; grouped charts include all intervals. CSV and copy exports contain every table row at the displayed precision. Potential spreadsheet formulas in category labels receive a protective apostrophe. CSV software may still reinterpret codes such as 001; import the category column as text to retain them.
No weights, two-way tables, workbook upload or saved datasets are provided. Data stays in memory until you clear it or leave the page. General site requests and any site-wide analytics are separate from this tool’s local calculations.
References: OpenStax on frequency tables and measurement levels; NIST on histograms; The Open University on frequency density.
Use it for a focused statistical question.
Count repeated measurements, summarize survey responses or organize a numerical distribution into intervals. Check the included observation count before sharing the table.
Frequency Table Maker questions
How the free browser tool behaves.
Can I use text categories?
Yes. Select text categories and use one label per line, or select a CSV column. Spaces within labels and leading-zero codes are preserved.
How is relative frequency calculated?
Divide the row’s count by all included observations. A count of 4 out of 20 gives 0.2, or 20%.
What does cumulative frequency tell me?
It gives the running count through an ordered row. For categories, supply and confirm a meaningful order first.
When should I group numerical values?
Group measurements when exact-value rows become hard to read. Keep an exact table when individual values are the point of the analysis.
How do I choose a class width?
Choose a width that matches the measurement scale and exposes useful variation. Compare it with the automatic suggestion, and check that the starting boundary covers the minimum.
Where does a value exactly on a boundary go?
It enters the interval starting at that boundary. The final upper boundary is the exception: it stays in the last interval.
Why might rounded percentages not total 100%?
Each percentage is rounded separately. Counts and unrounded shares remain unchanged; no row is adjusted to force the displayed sum.
Can I paste from Excel or Google Sheets?
Yes. Use one-per-line format for a single column or tab-separated format for multiple columns, then select the column to count. Upload CSV rather than an Excel workbook.
Is my dataset uploaded or saved?
No. This tool calculates locally and does not save observations in browser storage. Clear data removes the active inputs and results.
What if every number is the same?
An exact-value table has one row with frequency n. Automatic grouping cannot derive a width from a zero range; choose exact values or provide boundaries or a width.