What happens if a row has fewer parts than output columns?

Parts fill output columns left to right. Remaining output columns receive null.

What happens if a row has more parts than output columns?

Everything after the last split point is captured in the final output column. Add more output columns to separate each segment.

Does Split Columns modify the source column?

No. The source column is unchanged. Results are stored in the new named output columns.

How do I write a regex pattern for Split Columns?

Wrap the pattern in forward slashes: /your-pattern/. Each capture group maps to one output column in order. Invalid patterns block submission.

Can I split on a custom delimiter not in the dropdown?

Yes. Type any character or string directly into the delimiter dropdown to use it as a custom separator.

When should I use String Extract instead of Split Columns?

Use String Extract when you need only one value extracted from a column using regex. Split Columns is for dividing a column into multiple parts.

Split Columns | Edilitics Docs

Split a column into multiple new string columns by delimiter or regex. Define output column names and configure multiple split rules in one operation.

Split Columns divides a single column into multiple new string columns by separating values at a delimiter or regex pattern. You name each output column. The source column is unchanged. You can configure multiple split rules in one operation, each targeting a different source column.

When to Use Split Columns

Composite ID decomposition. Split ORD-2024-0001 on hyphen to extract order_prefix, order_year, and order_number as separate columns.
Full name to first and last. Split a full_name column on space to produce first_name and last_name.
Date string parsing. Split 2024-01-15 on hyphen to produce year, month, day columns without a cast operation.
Email decomposition. Use regex /^([^@]+)@(.+)$/ on an email column to separate username and domain.
Key-value extraction. Use regex capture groups to pull specific parts of a structured string pattern.

Sample Dataset

The examples in this doc use the Edilitics sample orders dataset. Download it to follow along in your own workspace.

edilitics_sample_orders.csv

Sample orders dataset for hands-on practice · 500 rows

Download

Relevant columns for Split Columns examples:

Prop

Type

Split Modes

Splits the column value at each occurrence of the chosen character or string. The value is divided into at most as many parts as there are output columns. If a value produces fewer parts than expected, the remaining output columns receive null.

Delimiter options:

Option	Character
Space
Tab	`\t`
Pipe	`\|`
Hyphen	`-`
Underscore	`_`
Custom	type any character or string

The delimiter dropdown is a creatable select - type any character or string directly to use a custom separator.

Overflow behaviour: if the value has more parts than output columns, everything after the last split point is captured in the final output column. "ORD-2024-0001" split on hyphen into two columns gives part_1 = "ORD" and part_2 = "2024-0001".

Underflow behaviour: "Kim" split on space into two columns gives first_name = "Kim" and last_name = null.

Splits using a regular expression with capture groups. Each capture group maps to one output column in order.

Enter the pattern wrapped in forward slashes: /pattern/

Example: /^([^@]+)@(.+)$/ applied to contact@edilitics.com produces two groups: contact and edilitics.com.

If a group does not match for a row, that output column receives null. The pattern must be a valid regex or the rule cannot be submitted.

Use Regex when:

The separator varies in width (e.g. one or more spaces: /(\S+)\s+(.+)/)
You need to match on structure rather than a fixed character
You want to extract specific parts using positional capture groups

If you want to extract a single value using a regex pattern - not split into multiple columns - use String Extract instead. String Extract is purpose-built for single-group extraction.

How to Apply Split Columns

Open the Split Columns operation

In your Transform pipeline, click Add Operation and select Split Columns from the operation list.

Select the column to split

Choose a source column from the Select Column To Split dropdown. All column types are available - string, numeric, and datetime columns all appear. Non-string values are converted to string internally before splitting.

Choose the split mode and separator

Use the type selector to choose Delimiter or Regex.

Delimiter: select Space, Tab, Pipe, Hyphen, or Underscore from the dropdown, or type a custom separator directly.
Regex: type a pattern wrapped in forward slashes (e.g. /^([^@]+)@(.+)$/). Each capture group (...) maps to one output column in order.

Name the output columns

In the New Columns section, name each output column. Two name fields appear by default. Click Add Column to add more output columns.

Each name must:

Contain only letters, numbers, and underscores
Not start with __ or a number
Not already exist in the dataset or in another split rule configured in this operation

Remove a column slot with its remove icon. A minimum of two output columns is required.

Click Split Column

Click Split Column to submit this rule. It appears in the table above the configuration panel.

To configure another split on a different source column, click Add Split Rule and repeat the steps above.

Save the operation

Click Save & Preview. Edilitics applies all split rules and adds the new columns to the dataset. A success toast confirms: "Column split applied. Preview reflects the updated columns."

Save & Preview is disabled until at least one split rule has been submitted.

Verify in the preview

Check the new columns. Look for unexpected nulls - these indicate rows where the value had fewer parts than output columns, or where a regex group did not match.

Before and After

Split rule: order_id on Hyphen delimiter into order_prefix, order_year, order_number.

Input:

order_id	region	revenue
ORD-2024-0001	North America	12450.00
ORD-2024-0047	Europe	8320.50
ORD-2024-0312	Asia Pacific	3750.75

Output:

order_id	region	revenue	order_prefix	order_year	order_number
ORD-2024-0001	North America	12450.00	ORD	2024	0001
ORD-2024-0047	Europe	8320.50	ORD	2024	0047
ORD-2024-0312	Asia Pacific	3750.75	ORD	2024	0312

Source column order_id unchanged. Three new string columns added.

Code Equivalent

-- Delimiter split (PostgreSQL / BigQuery / Snowflake)
SELECT
  *,
  SPLIT_PART(order_id, '-', 1) AS order_prefix,
  SPLIT_PART(order_id, '-', 2) AS order_year,
  SPLIT_PART(order_id, '-', 3) AS order_number
FROM orders;

-- Regex capture groups (PostgreSQL)
SELECT
  *,
  (REGEXP_MATCH(email, '^([^@]+)@(.+)$'))[1] AS email_user,
  (REGEXP_MATCH(email, '^([^@]+)@(.+)$'))[2] AS email_domain
FROM orders;

import polars as pl

# Delimiter split - at most 3 parts, extras captured in last column
df = df.with_columns([
    pl.col("order_id").str.splitn("-", 3).list.get(0).alias("order_prefix"),
    pl.col("order_id").str.splitn("-", 3).list.get(1).alias("order_year"),
    pl.col("order_id").str.splitn("-", 3).list.get(2).alias("order_number"),
])

# Regex capture groups
df = df.with_columns([
    pl.col("email").str.extract(r"^([^@]+)@(.+)$", 1).alias("email_user"),
    pl.col("email").str.extract(r"^([^@]+)@(.+)$", 2).alias("email_domain"),
])

After Save & Preview, the pipeline shows a DQ delta badge on this step - green if the table score improved, red if it dropped. See Data Quality Scoring for how scores are calculated.

Split Columns

When to Use Split Columns

Sample Dataset

Split Modes

How to Apply Split Columns

Before and After

Code Equivalent

Operation Reference

Frequently Asked Questions

Next Steps

String Extract

Merge Columns

Cast Data Types

Drop / Rename Columns

On this page