Add Headers to Your Columns

A column is referenced by the name of the column, which can be inferred from the first row of data in your dataset.

When a dataset is loaded, the application inserts a few transform steps automatically. If the application can identify that the first row of data is likely to contain the column headers for the dataset, this row is promoted to be used as the first version of the names of each column.

In some cases, however, this auto-generation of column headers may not work as expected, or you may have chosen at import time to not detect the structure of the dataset.

This section describes how you can generate column headers from within the application.

If your data has a header row in row 1

If the initial transforms do not promote your first row of data to be the column headers, you can use the following transform to promote the first row of data to be the column headers:


In some cases, the first row of data might not contain the headers or might not contain all of them.

For example, you may have some columns that contain nested data, and the column headers may not be immediately accessible.

Tip: After you unnest data in one or more columns, the first row might contain column headers. You can apply the header transform to promote these new values to be the names of the columns. The other column headers should not be overwritten.

If your data has a header row after row 1

In some cases, data may be imported such that header information is stored in a row other than the first one in your dataset.


  1. Hover your mouse over the black dot to the left of the row that contains your header information. The popup displays something similar to the following:

    Row 12
    Source Row 12
  2. Add a transform step using the source row number that you found:

    header sourcerownumber:12

    You can paste Wrangle steps into the Transform Builder.

  3. Add it to your recipe.

For more information, see Header Transform.

Send feedback about...

Google Cloud Dataprep Documentation