---
title: "stratifiedSplit(proportions:seed:annotationColumn:)"
framework: createml
role: symbol
role_heading: Instance Method
path: "createml/mlobjectdetector/datasource/stratifiedsplit(proportions:seed:annotationcolumn:)"
---

# stratifiedSplit(proportions:seed:annotationColumn:)

Generates a new data table by splitting the data source using the specified proportions.

## Declaration

```swift
func stratifiedSplit(proportions: [Double], seed: Int = timestampSeed(), annotationColumn: String) throws -> MLDataTable
```

## Parameters

- `proportions`: An array of doubles, each representing a portion of the data source. If these values don’t add up to 1.0, the method normalizes the numbers so that they do.
- `seed`: The value the method uses to initialize the random-number generator, which affects how the method splits the data.
- `annotationColumn`: The name of the column the method uses to split the data.

## Return Value

Return Value An MLDataTable containing the data source’s split contents.