Statistics Calculator

Percentile Calculator

Find thresholds. Explain ranks.

Enter your data

Quantiles and tie-aware ranks.

Dataset values

Separate signed decimals, scientific e notation or integer fractions with commas, spaces, semicolons, pipes or new lines. Decimal point only; commas separate values, never thousands.

Input method

Each method keeps its own data. Frequency counts represent repeated observations, not arbitrary weights.

Conventions and display

Unit label

A literal label only; all observations must already use the same measurement unit. No unit conversion is performed.

Requested percentile

Enter a percentage from zero through one hundred, without a percent sign. Fractions and scientific e notation are accepted.

Quantile method

Type 7 uses inclusive linear interpolation. All nine methods clamp outside contributing ranks to the nearest endpoint; Type 6 is not Excel PERCENTILE.EXC.

Score to rank

A value in the same unit as your data, evaluated against all repetition counts. The score does not have to occur in the dataset.

Percentile rank convention

Changes the highlighted rank. All four conventions are shown for comparison, including their treatment of tied observations.

Batch percentiles

Optional: up to twenty-five percentages using the selected quantile method. Separate with commas, spaces, semicolons or line breaks. Clear to omit the batch.

Rounded decimal places

Exact rational answers do not change. Tiny nonzero views use scientific notation. Halfway values round away from zero.

Import a local CSV or text file

Imports into the selected method. Files must be UTF-8, at most 64 KiB, and contain only supported numeric data plus an optional frequency header. Invalid files leave current data unchanged.

Worked datasets

Show step-by-step solutionShow sorted rank tableCompare all nine quantile methods

Supports up to $2{,}000$ source observations or frequency rows, and total count $n\leq10^9$ . Nonzero source magnitudes range from $10^{-100}$ through $10^{100}$ , with forty significant digits per number. Combined exact arithmetic is bounded to $1{,}500$ numerator or denominator digits. Results describe the supplied data; no missing values or outliers are removed automatically.

5 observations; 4 distinct exact values. All active source values retained.

Percentile value

Type 7 — Inclusive linear

Q\!\left(75\%\right)=6

$\approx 6$ points

A percentile value is a threshold in the original measurement unit. The percentile rank below answers a different question about the evaluated score.

Percentile rank of your score

The score remains separate from the percentile threshold. Ties create different defensible rank percentages; choose the convention required by your report.

Mean: half of ties

v=4,\quad R=40\%

Observation count: $5$
Below score: $1$
Equal to score: $2$
Above score: $2$

Strict: below only

20

\approx 20

percent

Weak: at or below

60

\approx 60

percent

Mean: half of ties

40

\approx 40

percent

Rank: average matching rank

50

\approx 50

percent

These percentages are empirical ranks within the supplied data, not exam grades, probabilities of future outcomes or fitted normal-distribution percentiles. Average matching rank differs from the half-ties rule when the score occurs in the dataset.

Compare all nine quantile methods

All methods use the same exact sorted observations and requested percentage. Endpoint ranks are clamped to the observed minimum or maximum. Type 6 here follows that rule and does not reproduce Excel exclusive-function errors.

Type 1 — Nearest rank

6

\approx 6

points

Type 2 — Averaged inverse CDF

6

\approx 6

points

Type 3 — Nearest even rank

6

\approx 6

points

Type 4 — Linear empirical CDF

\frac{11}{2}

\approx 5.5

points

Type 5 — Hazen

\frac{27}{4}

\approx 6.75

points

Type 6 — Weibull (clamped)

\frac{15}{2}

\approx 7.5

points

Type 7 — Inclusive linear

6

\approx 6

points

Type 8 — Median-unbiased

7

\approx 7

points

Type 9 — Normal-unbiased

\frac{111}{16}

\approx 6.9375

points

Batch percentile values

Every requested percentage uses Type 7 — Inclusive linear. Duplicate requests remain in their original order.

Percentile	Exact threshold and rounded view
$0\%$	$2$ $\approx 2$ points
$10\%$	$\frac{14}{5}$ $\approx 2.8$ points
$25\%$	$4$ $\approx 4$ points
$50\%$	$4$ $\approx 4$ points
$75\%$	$6$ $\approx 6$ points
$90\%$	$\frac{39}{5}$ $\approx 7.8$ points
$100\%$	$9$ $\approx 9$ points

Step-by-step percentile solution

Ordered observations are written $x_{(k)}$ . The quantile calculation uses the probability $p$ , rank floor $j$ and interpolation weight $\gamma$ . For the score calculation, $B$ , $E$ and $A$ count observations below, equal to and above the score.

1. Validate, sort and count
Every active token is checked. Equivalent fractions and decimals combine into exact frequency rows; zero-frequency rows are explicitly excluded.
$n=\sum_i f_i=5,\quad u=4$
2. Convert the requested percentile to a probability
Type 7 — Inclusive linear. The requested number is a percentage, not a probability input.
$p=\frac{75}{100}=\frac{3}{4}$
3. Locate the method-specific rank
The raw one-based interpolation position is split into an integer floor and fractional part. Types one through three apply their discontinuous selection rules instead of ordinary linear interpolation.
$h=np+m=4,\quad j=\lfloor h\rfloor=4,\quad \gamma=0$
4. Select or interpolate adjacent order statistics
Ranks outside the observed range repeat the nearest endpoint, consistently with the nine clamped quantile definitions. These are not the domain restrictions of Excel PERCENTILE.EXC.
$Q_{7}\!\left(\frac{3}{4}\right)=\left(1-0\right)\left(6\right)+0\left(9\right)=6$
5. Count values relative to the score
Below, equal and above are exact frequency totals, not counts of distinct rows. No normal-distribution model is fitted.
$v=4,\quad B=1,\quad E=2,\quad A=2,\quad B+E+A=5$
6. Apply the selected tie convention
Mean: half of ties. Average matching rank includes a half-rank correction only when the score occurs in the dataset. Without a match, all four conventions agree.
$R_{\mathrm{mean}}=100\frac{B+E/2}{n}=40\%$

Sorted value and rank audit

Each distinct value occupies a closed interval of one-based observation ranks. Strict and weak percentages mark the empirical distribution immediately before and after its ties.

Showing distinct rows 1–4 of 4. Printing includes this page; copying and CSV include every row.

Observed value	Frequency	First rank	Last rank	Strict percent	Weak percent
$2$ $\approx 2$ points	$1$	$1$	$1$	$0$ $\approx 0$	$20$ $\approx 20$
$4$ $\approx 4$ points	$2$	$2$	$3$	$20$ $\approx 20$	$60$ $\approx 60$
$6$ $\approx 6$ points	$1$	$4$	$4$	$60$ $\approx 60$	$80$ $\approx 80$
$9$ $\approx 9$ points	$1$	$5$	$5$	$80$ $\approx 80$	$100$ $\approx 100$

What Can You Create?

Find thresholds and understand score positions

Calculate a measurement threshold

Choose a requested percentage and inspect its value in the original measurement unit. The selected quantile type determines whether the answer selects an observed value, averages a boundary pair or interpolates between observations. Exact rank positions stay visible so a result between source values can be explained without pretending that the interpolated measurement occurred in the dataset.

Compare scores with their reference group

Enter a separate score and inspect how many observations fall below it, match it and exceed it. The four tie conventions make the effect of repeated scores explicit. Keep the reference group and convention with the result when discussing assessment data or ordered measurements. A score's relative position can change when the reference dataset changes even though the score itself stays fixed.

Build a batch threshold report

Request several percentages in one calculation while retaining a single quantile convention. The batch preserves the order of requests and keeps duplicates rather than silently reorganizing your report. Frequency input represents repeated observations compactly. Copying and CSV preserve the complete sorted rank audit, allowing someone else to trace each threshold back to its contributing observation ranks.

Percentile formulas and step-by-step examples

Sort the observations as $x_{(1)}\leq\cdots\leq x_{(n)}$ . Convert the requested percentage $P$ to $p=P/100$ . Define $h=np+m$ , $j=\lfloor h\rfloor$ and $g=h-j$ . The method supplies $m$ and the interpolation weight $\gamma$ , producing the following threshold. For a rank below the first or above the last observation, repeat the nearest endpoint.

Q(p)=(1-\gamma)x_{(j)}+\gamma x_{(j+1)}

Type	Offset	Interpolation rule
$1$	$m=0$	$\gamma=\begin{cases}0&g=0\\1&g>0\end{cases}$
$2$	$m=0$	$\gamma=\begin{cases}1/2&g=0\\1&g>0\end{cases}$
$3$	$m=-1/2$	$\gamma=\begin{cases}0&g=0\text{ and }j\text{ even}\\1&\text{otherwise}\end{cases}$
$4$	$m=0$	$\gamma=g$
$5$	$m=1/2$	$\gamma=g$
$6$	$m=p$	$\gamma=g$
$7$	$m=1-p$	$\gamma=g$
$8$	$m=(p+1)/3$	$\gamma=g$
$9$	$m=p/4+3/8$	$\gamma=g$

An interpolation example

For the ordered values $2,4,4,6,9$ , request $P=40$ using Type 7. The rank is $h=1+(5-1)(2/5)=13/5$ . Its floor is $j=2$ and its fractional part is $g=3/5$ . Both contributing observations equal $4$ , so interpolation returns $4$ . Under Type 2, the exact boundary $np=2$ also averages the second and third observations, while other datasets can make these two methods disagree.

Q_7(2/5)=(1-3/5)\cdot4+(3/5)\cdot4=4

A tie example

In $1,2,3,3,4$ , evaluate the score $v=3$ . There are $B=2$ observations below it, $E=2$ matches and $A=1$ above it. Strict and weak ranks describe opposite edges of the tie group. The mean convention places the score halfway through that group's cumulative jump; average matching rank averages the two occupied positions instead.

R_{\mathrm{strict}}=100B/n=40\%, \quad R_{\mathrm{weak}}=100(B+E)/n=80\%

R_{\mathrm{mean}}=100(B+E/2)/n=60\%, \quad R_{\mathrm{rank}}=100(B+(E+1)/2)/n=70\%

The last formula applies when matches exist. With no matching observation, the average-rank convention becomes $100B/n$ , so all four rank conventions agree. A larger-value-is-better interpretation is a context choice; these formulas always count in ascending numerical order.

Why Users Love This Tool

Why the method belongs beside the answer

Source values stay accountable

Every active token is validated rather than removed as a missing value.
Exact comparison keeps large, closely spaced measurements in the correct order.
Frequency counts represent repetitions and never allocate a billion-element list.
Source and query limits are reported explicitly rather than truncating a result.

Conventions stay visible

Nine quantile definitions share a clearly stated endpoint policy.
Four score-rank conventions expose the effect of ties.
Batch thresholds keep the same method throughout the report.
Exact results remain separate from independently rounded numerical views.

Perfect For

For lessons, ordered measurements and reproducible reports

Students tracing an ordered rank

Move from an unsorted numerical list to explicit one-based rank intervals. Follow a fractional position into the two contributing observations and its interpolation weight. Compare a nearest-rank answer with a linear threshold, then change to tied values to see why a rank percentage needs its own definition. The examples make boundary behavior visible without relying on a hidden software default.

Teachers comparing definitions

Use the same source values to compare established quantile definitions without changing the reference dataset between examples. Singleton, constant and even-count presets help isolate endpoint and interpolation behavior. Score counts connect a cumulative distribution interpretation to the separate average matching-rank rule. Exact working supports a worksheet explanation while the batch report provides several checkpoints under a chosen convention.

People reviewing numerical reports

Import a supported local numerical file, retain repetition counts and compare the resulting thresholds with another documented method. Keep the source-unit label and reference dataset with your exported report. This calculator describes supplied observations; it does not establish representativeness, remove outliers or infer unknown measurements from grouped intervals. Check the provenance of the source before extending a rank statement to a wider population.

How It Works

How it works in three quick steps.

Enter observations and choose your questions

Paste a numerical list or select a value and frequency table. Enter the requested percentile as a percentage and a separate score whose percentile rank you want to evaluate. The score can fall between observations. Check the unit, separators and source count before comparing results. Local UTF-8 CSV or text files can populate the active editor without uploading the dataset.

Select quantile and tie conventions

Choose an explicit quantile type for the threshold and a tie-counting rule for the score rank. Compare the nine threshold methods and the four score-rank conventions on the same source data. Inspect the ordered ranks and exact interpolation weight in the working. Batch requests use the selected quantile method rather than switching conventions automatically at the median or quartiles.

Audit the ranks and save the record

Follow each distinct observation through its frequency and inclusive rank interval. Copy or download the complete report to retain all rows and methods, including hidden sections. Share a validated active dataset when it fits the link limit. Use browser printing to save the visible calculation and current audit page as PDF, keeping the complete CSV for longer datasets and exact expressions.

Download & Print

Keep the source, conventions and complete rank audit

Copy a readable calculation

Copy the two-direction result with exact fractions, numerical views, ordered steps, comparison methods and every frequency row. The summary converts visual mathematical typesetting to readable plain text for destinations that do not support LaTeX. Hidden sections are retained, so a compact screen view does not silently become an incomplete saved calculation. Keep the method names attached when quoting any threshold or rank.

Download the full CSV

The rectangular CSV includes active source text, unit label, input mode, selected conventions, all nine method certificates, all four rank percentages, batch queries and complete frequency intervals. Formula-like cells are protected for spreadsheet use. It is an output report rather than an input-editor template; to import a source file, supply only numeric observations or the documented value and frequency columns.

Save a focused PDF worksheet

Choose Print / Save PDF and select your browser's PDF destination. The result and visible working remain while editing controls and supporting article sections are removed. Printing includes the current audit page, not an expanded full dataset. Long exact expressions carry a CSV notice rather than being clipped. Use the complete download when you need every source row or an unusually long fraction.

FAQ

Frequently Asked Questions

What is the difference between a percentile value and a percentile rank?

A percentile value is a measurement threshold obtained from an ordered dataset and a requested percentage. A percentile rank instead starts with a score and asks how that score compares with the observations. The threshold has the original measurement unit; the rank is a percentage. They are not guaranteed to reverse each other, especially with ties, a small dataset or an interpolated threshold that is absent from the source observations.

Which percentile method should I choose?

Follow the convention specified by your course, report or receiving software. The default is inclusive linear interpolation, Type 7. Nearest rank, Type 1, selects an observed value instead. This calculator exposes nine established quantile definitions and compares their results rather than treating one answer as universally correct. Record the method beside the result, and compare it with your reference before interpreting a small numerical difference as an error.

How are tied scores handled in percentile ranks?

Strict rank counts only observations below the score. Weak rank includes all equal observations. The mean convention includes half of the equal observations. Average matching rank averages the one-based ranks of matches and expresses that position as a percentage of the count. For

1,2,3,3,4

evaluated at

3

, those percentages are

40\%

80\%

60\%

and

70\%

, respectively. A report must identify which convention it uses.

Does Type 6 reproduce Excel PERCENTILE.EXC?

Type 6 uses the Weibull interpolation position, but this tool repeats the nearest observed endpoint when a contributing rank lies outside the dataset. That endpoint policy is consistent across the nine offered quantile types. Excel's exclusive function has additional domain restrictions, so identical rank formulas do not imply identical error behavior. Do not use this Type 6 option as a promise of exclusive-spreadsheet compatibility at extreme probabilities or for a singleton.

What happens at endpoints, with one observation or with constant data?

The endpoint percentages return the observed minimum and maximum under all nine supported methods. A singleton returns its only value for every requested percentile, and a constant dataset returns its shared value throughout. Percentile ranks still depend on the score and tie convention: at a singleton's observed value, strict rank is

0\%

, weak rank is

100\%

, mean rank is

50\%

and average matching rank is

100\%

. These are valid conventions, not missing results.

Can I use frequencies instead of pasting repeated values?

Yes. Each frequency is a nonnegative whole-number repetition count, and the total count controls the order-statistic positions. Equivalent values are grouped exactly, while zero-count rows are reported as excluded. The tool looks up ranks within cumulative intervals rather than expanding a large list in memory. Arbitrary reliability weights and grouped class intervals are different input models and are not inferred from the frequency editor or a measurement label.

What numbers, precision and file formats are supported?

Use finite signed decimals, scientific e notation or integer fractions. A decimal point is required; commas separate values instead of thousands. The source limit is two thousand observations or frequency rows, with total frequency at most one billion. Each editor is bounded to sixty-five thousand five hundred thirty-six characters and a local file to sixty-four KiB. Precision changes the rounded view only. Exact fractions remain available, and unsupported tokens are rejected instead of being silently dropped.

Can I use these ranks as exam scores or normal-distribution probabilities?

An empirical percentile rank compares a score only with the supplied observations. It is not the percentage of questions answered correctly, a guarantee about a future score or a probability from a fitted normal distribution. An external testing organization may use another norm group or tie convention. Verify that population and method before comparing a published rank with this calculation. Use a separate distribution model when your question explicitly requires one.

What do sharing, CSV and PDF preserve?

A share link retains the active input and validated calculation choices within a bounded link size. The inactive editor is omitted. Copying and CSV retain every sorted row, all nine comparisons, all rank conventions, batch results and ordered working even if those sections are hidden. CSV also includes the active source text and protects formula-like cells. Printing keeps currently visible results and the selected audit page; long exact expressions point to their complete CSV record.

Related Calculators

Related calculators

Mean Median Mode Calculator

Review numerical centers, repeated values and quartile conventions.

Try it

Standard Deviation Calculator

Measure spread using sample and population variance with an exact audit.

Try it

Z-Score Calculator

Compare standardized distance from a mean when that model fits your question.

Try it

About This Tool

About this percentile and percentile rank calculator

Toolarithm's Percentile Calculator brings two commonly confused questions into one traceable workflow. One question asks for a threshold at a requested percentage in an ordered dataset. The other asks for the relative position of a supplied score. The interface retains both questions, names the quantile and tie conventions separately, and shows the counts and ranks behind the answers. This matters when a small change in a software default produces a different published threshold or when repeated scores make a single unqualified rank percentage ambiguous.

The numerical engine treats supported observations as exact fractions. Sorting and interpolation therefore preserve the entered numbers instead of first rounding them to binary floating-point values. A frequency table counts repeated observations through cumulative intervals without expanding the source. This supports large multiplicities while keeping source-size limits explicit. Scientific notation and integer fractions are accepted, but missing observations, arbitrary text, grouped intervals and probability weights are not guessed. A measurement label is literal context and does not perform a unit conversion.

The educational material connects each result to the selected definition and its endpoint policy. Exact certificates, rounded views and complete sorted rows can be retained through copying or CSV, while browser printing creates a focused visible worksheet. These outputs support checking a calculation; they do not certify that a dataset is representative or that a rank has the same meaning as an exam grade. When comparing with an external report, match its reference population, quantile definition, treatment of ties and rounding policy before drawing conclusions from an apparent disagreement.

Editorial Transparency

Who maintains this page

Ownership and review

Written and maintained by the Toolarithm editorial team. No review date is shown without a maintained editorial record. No independent professional review is claimed.

Methodology

Equations are checked by substituting worked examples into the stated relationship and comparing the result with the interactive implementation.

Read the editorial methodology

Dates and sources

Review dates change only after a substantive method or content check.

Keep building

Explore more calculators

Calculators hub

Percentile Calculator

Enter your data

Conventions and display

Percentile value

Percentile rank of your score

Strict: below only

Weak: at or below

Mean: half of ties

Rank: average matching rank

Compare all nine quantile methods

Type 1 — Nearest rank

Type 2 — Averaged inverse CDF

Type 3 — Nearest even rank

Type 4 — Linear empirical CDF

Type 5 — Hazen

Type 6 — Weibull (clamped)

Type 7 — Inclusive linear

Type 8 — Median-unbiased

Type 9 — Normal-unbiased

Batch percentile values

Step-by-step percentile solution

1. Validate, sort and count

2. Convert the requested percentile to a probability

3. Locate the method-specific rank

4. Select or interpolate adjacent order statistics

5. Count values relative to the score

6. Apply the selected tie convention

Sorted value and rank audit

Find thresholds and understand score positions

Calculate a measurement threshold

Compare scores with their reference group

Build a batch threshold report

Percentile formulas and step-by-step examples

An interpolation example

A tie example

Why the method belongs beside the answer

Source values stay accountable

Conventions stay visible

For lessons, ordered measurements and reproducible reports

Students tracing an ordered rank

Teachers comparing definitions

People reviewing numerical reports

How it works in three quick steps.

Enter observations and choose your questions

Select quantile and tie conventions

Audit the ranks and save the record

Keep the source, conventions and complete rank audit

Copy a readable calculation

Download the full CSV

Save a focused PDF worksheet

Frequently Asked Questions

Related calculators

Mean Median Mode Calculator

Standard Deviation Calculator

Z-Score Calculator

About this percentile and percentile rank calculator

Who maintains this page

Ownership and review

Methodology

Dates and sources

Explore more calculators