General information

This brick is designed to let the user change columns' types in the input dataset.


Brick Location

Data Manipulation → Convert / Replace → Typization

Brick Parameters

  • Column
    • The name of the column in the input data frame, which type we want to change.
  • Type
    • New type for the specified column.
      Here we support boolean, integer, float, string, category and datetime types.
  • Date format
    • Additional setting for casting into datetime format.
      This field is enabled only if new type is datetime.
  • NaN fraction
    • Float and datetime types have an option to choose a NaN fraction of invalid values. This determines the percentage of invalid data that can be tolerated without failing the conversion. This setting can be helpful in case there is the desired threshold of possibly corrupted data that should not be exceeded. The default behavior is set to 100 meaning that 100% of data can fail to convert resulting in NaN values.

Brick Inputs/Outputs

  • Inputs
    • Brick takes the data set without any restrictions.
  • Outputs
    • Brick produces the result as a new dataset with new columns' types.