Share via

Convert to Indicator Values


Support for Machine Learning Studio (classic) will end on 31 August 2024. We recommend you transition to Azure Machine Learning by that date.

Beginning 1 December 2021, you will not be able to create new Machine Learning Studio (classic) resources. Through 31 August 2024, you can continue to use the existing Machine Learning Studio (classic) resources.

ML Studio (classic) documentation is being retired and may not be updated in the future.

Converts categorical values in columns to indicator values

Category: Data Transformation / Manipulation


Applies to: Machine Learning Studio (classic) only

Similar drag-and-drop modules are available in Azure Machine Learning designer.

Module overview

This article describes how to use the Convert to Indicator Values module in Machine Learning Studio (classic). The purpose of this module is to convert columns that contain categorical values into a series of binary indicator columns that can more easily be used as features in a machine learning model.

How to configure Convert to Indicator Values

  1. Add the Convert to Indicator Values module to your Machine Learning experiment, and connect it to the dataset containing the columns you want to convert. You can find this module under Data Transformations, in the Manipulation category.

  2. Use the Column Selector to choose one or more categorical columns.

    To ensure that the columns you select are categorical, use Edit Metadata before Convert to Indicator Values in your experiment, to mark the target column as categorical.

  3. Select the Overwrite categorical columns option if you want to output only the new Boolean columns.

    By default, this option is off, which lets you see the categorical column that is the source, together with the related indicator columns.


    If you choose the option to overwrite, the source column is not actually deleted or modified. Instead, the new columns are generated and presented in the output dataset, and the source column remains available in the workspace. If you need to see the original data, you can use the Add Columns module at any time to add the source column back in.

  4. Run the experiment.


For example, suppose you have a column with scores that indicate whether a server has a high, medium or low probability of failure.

Server ID Failure score
10301 Low
10302 Medium
10303 High

When you apply Convert to Indicator Values, the single column of labels is converted into multiple columns containing Boolean values:

Server ID Failure score - Low Failure score - Medium Failure score - High
10301 1 0 0
10302 0 1 0
10303 0 0 1

Here is how the conversion works:

  • In the Failure score column that describes risk, there are only three possible values (High, Medium, and Low), and no missing values. Therefore exactly three new columns are created.

  • The new indicator columns are named based on the column headings and values of the source column, using this pattern: <source column>- <data value>.

  • There should be a 1 in exactly one indicator column, and 0 in all other indicator columns. That is because each server can have only one risk rating.

You can now use the three indicator columns as features and analyze their correlation with other properties that are associated with different risk level.


To see examples of how this module is used, see the Azure AI Gallery:

  • Breast cancer detection: Patients are binned into groups based on patient ID numbers, and then Indicator Values is used to flag which group the patient belongs to. Later, the group indicators are used when scoring models.

  • Direct marketing: Probabilities are compared to a constant by using Apply Math Operation, and the Yes/No values that indicate whether the score was above or below the constant are turned into new indicator columns.

  • Network intrusion detection: Log data is loaded from Azure storage. The class variable (which describes, for example, if an attack is a rootkit or buffer overflow) is converted to a categorical column and then expanded to multiple indicator values.

Technical notes

This section contains implementation details, tips, and answers to frequently asked questions.

Usage tips

  • Only columns that are marked as categorical can be converted to indicator columns. If you see this error, it is likely that one of the columns you selected is not categorical:

    Error 0056: Column with name <column name> is not in an allowed category.

    By default most string columns are handled as string features, so you must explicitly mark them as categorical using Edit Metadata.

  • An error is displayed if you do not select at least one categorical column.

  • There is no limit on the number of columns that you can convert to indicator columns. However, because each column of values can yield multiple indicator columns, you might want to convert and review just a few columns at a time.

  • If the column contains missing values, a separate indicator column is created for the missing category, with this name: <source column>- Missing

  • If the column that you convert to indicator values contains numbers, they must be marked as categorical like any other feature column. After you have done so, the numbers are treated as discrete values. For example, if you have a numeric column with MPG values ranging from 25 to 30, a new indicator column would be created for each discrete value:

    Make Highway mpg -25 Highway mpg -26 Highway mpg -27 Highway mpg -28 Highway mpg -29 Highway mpg -30
    Alfa Romeo 0 0 0 0 0 1

    To avoid getting a huge number of indicator columns, we recommend that you first check the number of values in the column, and bin or quantize the data appropriately.

Expected inputs

Name Type Description
Dataset Data Table Dataset with categorical columns

Module parameters

Name Range Type Default Description
Categorical columns to convert Any ColumnSelection Select categorical columns to convert to indicator matrices.
Overwrite categorical columns Any Boolean false If True, overwrite the selected categorical columns; otherwise, append the resulting indicator matrices to the dataset.


Name Type Description
Results dataset Data Table Dataset with categorical columns converted to indicator matrices.

See also

Data Transformation
A-Z Module List