$percentile
Definition
Computes single-document statistics, ignoring nonnumeric values.
Expression operator · Statistics · Studio Visual: available. Studio validates the complete expression separately.
Syntax
{
"$percentile": {
"input": <input>,
"p": <p>,
"method": <method>
}
}
Replace angle-bracket placeholders with expressions or values.
Optional parameters may be omitted.
Parameters
| Parameter | Required | Description |
|---|---|---|
input |
Yes | Expression supplying the value to process. A field reference such as "$price" reads from the current document. |
p |
Yes | Array of requested percentiles, each between 0 and 1 inclusive. |
method |
Yes | Literal "approximate". Other method names are rejected. |
Supported values
method must be the literal "approximate". See the results below for the engine’s numeric behavior.
Behavior
The following observations use the example expression below. Each row substitutes one argument while retaining the other example arguments. A missing field differs from an explicit null; the result shown is what Core actually returned.
For the input argument:
| Input case | Supplied value | Result |
|---|---|---|
| null | null | [null,null,null] |
| missing | Field absent | [null,null,null] |
| decimal | 2.5 | [2.5,2.5,2.5] |
| numeric string | "12" | [null,null,null] |
| text | "hello" | [null,null,null] |
| boolean | true | [null,null,null] |
| typed date | {"$date":"2026-01-15T15:00:00Z"} | [null,null,null] |
| empty array | [] | [null,null,null] |
| numeric array | [1,2,3] | [1.0,2.0,3.0] |
| object | {"x":1} | [null,null,null] |
Argument handling
| Invocation | Result |
|---|---|
| {"$percentile":[]} | Error — $percentile requires [method, p, input] |
| {"$percentile":{"unexpected":1}} | Error — $percentile requires [method, p, input] |
Examples
Expression:
{
"$percentile": {
"input": [
10,
20,
30,
40
],
"p": [
0,
0.5,
1
],
"method": "approximate"
}
}
Input document:
{}
Observed SrotaX result: [10.0,20.0,40.0].
Detailed input cases
Expand the complete argument-by-argument test matrix
Each table changes only the indicated argument from the example above. Other arguments retain their baseline values. Values normally come from the input field probe; literal configuration arguments are replaced directly. “Missing” means the field is absent, not explicitly null. Numeric strings are strings, not numbers.
Argument: input
| Input case | Supplied value | SrotaX result |
|---|---|---|
| null | null | [null,null,null] |
| missing | Field absent | [null,null,null] |
| zero | 0 | [0.0,0.0,0.0] |
| negative | -2 | [-2.0,-2.0,-2.0] |
| decimal | 2.5 | [2.5,2.5,2.5] |
| numeric string | "12" | [null,null,null] |
| text | "hello" | [null,null,null] |
| empty string | "" | [null,null,null] |
| boolean | true | [null,null,null] |
| false | false | [null,null,null] |
| typed date | {"$date":"2026-01-15T15:00:00Z"} | [null,null,null] |
| decimal128 | {"$numberDecimal":"1.25"} | [1.25,1.25,1.25] |
| empty array | [] | [null,null,null] |
| numeric array | [1,2,3] | [1.0,2.0,3.0] |
| mixed array | [1,null,"x",3] | [1.0,1.0,3.0] |
| nested array | [[1,2],[3]] | [null,null,null] |
| empty object | {} | [null,null,null] |
| object | {"x":1} | [null,null,null] |
| ISO date string | "2026-01-15T15:00:00Z" | [null,null,null] |
| omitted named argument | Argument removed | Error — $percentile requires [method, p, input] |
Argument: p
| Input case | Supplied value | SrotaX result |
|---|---|---|
| null | null | Error — $percentile p must be an array |
| zero | 0 | Error — $percentile p must be an array |
| negative | -2 | Error — $percentile p must be an array |
| decimal | 2.5 | Error — $percentile p must be an array |
| numeric string | "12" | Error — $percentile p must be an array |
| text | "hello" | Error — $percentile p must be an array |
| empty string | "" | Error — $percentile p must be an array |
| boolean | true | Error — $percentile p must be an array |
| false | false | Error — $percentile p must be an array |
| typed date | {"$date":"2026-01-15T15:00:00Z"} | Error — $percentile p must be an array |
| decimal128 | {"$numberDecimal":"1.25"} | Error — $percentile p must be an array |
| empty array | [] | Error — $percentile p must be an array |
| numeric array | [1,2,3] | Error — $percentile p values must be numbers from 0 to 1 |
| mixed array | [1,null,"x",3] | Error — $percentile p values must be numbers from 0 to 1 |
| nested array | [[1,2],[3]] | Error — $percentile p values must be numbers from 0 to 1 |
| empty object | {} | Error — $percentile p must be an array |
| object | {"x":1} | Error — $percentile p must be an array |
| ISO date string | "2026-01-15T15:00:00Z" | Error — $percentile p must be an array |
| omitted named argument | Argument removed | Error — $percentile requires [method, p, input] |
Argument: method
| Input case | Supplied value | SrotaX result |
|---|---|---|
| null | null | Error — $percentile method must be the literal string approximate |
| zero | 0 | Error — $percentile method must be the literal string approximate |
| negative | -2 | Error — $percentile method must be the literal string approximate |
| decimal | 2.5 | Error — $percentile method must be the literal string approximate |
| numeric string | "12" | Error — $percentile method must be the literal string approximate |
| text | "hello" | Error — $percentile method must be the literal string approximate |
| empty string | "" | Error — $percentile method must be the literal string approximate |
| boolean | true | Error — $percentile method must be the literal string approximate |
| false | false | Error — $percentile method must be the literal string approximate |
| typed date | {"$date":"2026-01-15T15:00:00Z"} | Error — $percentile method must be the literal string approximate |
| decimal128 | {"$numberDecimal":"1.25"} | Error — $percentile method must be the literal string approximate |
| empty array | [] | Error — $percentile method must be the literal string approximate |
| numeric array | [1,2,3] | Error — $percentile method must be the literal string approximate |
| mixed array | [1,null,"x",3] | Error — $percentile method must be the literal string approximate |
| nested array | [[1,2],[3]] | Error — $percentile method must be the literal string approximate |
| empty object | {} | Error — $percentile method must be the literal string approximate |
| object | {"x":1} | Error — $percentile method must be the literal string approximate |
| ISO date string | "2026-01-15T15:00:00Z" | Error — $percentile method must be the literal string approximate |
| omitted named argument | Argument removed | Error — $percentile requires [method, p, input] |
Additional boundary examples
Show 25 tested examples
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
10,
20,
30
]
}
SrotaX: [30.0,10.0,10.0,20.0,30.0,20.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
10,
20,
30,
40
]
}
SrotaX: [40.0,10.0,10.0,20.0,30.0,20.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": []
}
SrotaX: [null,null,null,null,null,null].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": null
}
SrotaX: [null,null,null,null,null,null].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": 5
}
SrotaX: [5.0,5.0,5.0,5.0,5.0,5.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
1,
"x",
null,
3
]
}
SrotaX: [3.0,1.0,1.0,1.0,3.0,1.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
1,
[
50
],
3
]
}
SrotaX: [3.0,1.0,1.0,1.0,3.0,1.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
"x",
null,
{}
]
}
SrotaX: [null,null,null,null,null,null].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
1,
{
"$numberDouble": "NaN"
},
3
]
}
SrotaX: [3.0,1.0,1.0,1.0,3.0,1.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
1,
{
"$numberDouble": "Infinity"
},
3
]
}
SrotaX: [Infinity,1.0,1.0,3.0,Infinity,3.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
{
"$numberDecimal": "1.2"
},
2,
3
]
}
SrotaX: [3.0,1.2,1.2,2.0,3.0,2.0].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
1e+100,
2e+100,
3e+100
]
}
SrotaX: [3e+100,1e+100,1e+100,2e+100,3e+100,2e+100].
Expression:
{
"$percentile": {
"input": "$values",
"method": "approximate",
"p": [
1,
0,
0.25,
0.5,
0.75,
0.5
]
}
}
Input:
{
"values": [
{
"$numberDouble": "Infinity"
}
]
}
SrotaX: [Infinity,Infinity,Infinity,Infinity,Infinity,Infinity].
Expression:
{
"$percentile": {
"input": "$missing",
"method": "approximate",
"p": [
0,
0.5,
1
]
}
}
Input:
{}
SrotaX: [null,null,null].
Expression:
{
"$percentile": {
"input": [
1,
2,
3
],
"method": "continuous",
"p": [
0.5
]
}
}
Input:
{}
SrotaX: Error — $percentile method must be the literal string approximate.
Expression:
{
"$percentile": {
"input": [
1,
2,
3
],
"method": null,
"p": [
0.5
]
}
}
Input:
{}
SrotaX: Error — $percentile method must be the literal string approximate.
Expression:
{
"$percentile": {
"input": [
1,
2,
3
],
"method": {
"$literal": "approximate"
},
"p": [
0.5
]
}
}
Input:
{}
SrotaX: Error — $percentile method must be the literal string approximate.
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": []
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: Error — $percentile p must be an array.
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": [
-1
]
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: Error — $percentile p values must be numbers from 0 to 1.
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": [
1.1
]
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: Error — $percentile p values must be numbers from 0 to 1.
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": [
"0.5"
]
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: Error — $percentile p values must be numbers from 0 to 1.
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": [
null
]
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: Error — $percentile p values must be numbers from 0 to 1.
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": {
"$literal": [
0.5
]
}
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: [2.0].
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": "$ps"
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: Error — $percentile p must be constant.
Expression:
{
"$percentile": {
"input": [
1,
2,
3,
4
],
"method": "approximate",
"p": [
{
"$divide": [
1,
2
]
}
]
}
}
Input:
{
"ps": [
0.5
]
}
SrotaX: [2.0].
Verification
This page records 86 executed cases. It covers individual argument substitutions, not every cross-product of arguments, locale, daylight-saving transitions, numeric overflow, collation or resource limits. No untested behavior is guaranteed.
Download exact expressions, inputs and SrotaX results. Errors are observed messages, not stable public error codes. The reference generator reads the checked-in execution report; do not edit generated observations by hand.
MongoDB differences
Compared with MongoDB 9.0.4: 36 exact BSON matches, 50 both rejected, 0 value/type/acceptance differences, 0 object-field-order differences.
“Exact” includes numeric BSON types and object key order. A value of 3 as int32 differs from 3.0 as double. Object-field-order differences are reported separately; object key ordering is not enforced by the engine regression baseline. Both rejected does not mean the error codes or messages match. This is finite-case evidence, not full compatibility certification.
No differences were observed in the compared cases.