Take a collection of integers or missing values:

By default, integers will be interpreted using the "Integer64" type in this machine:

Construct three columns with the same data but casting to different types of integers:

Check the stored types:

Take a collection of Boolean or missing values:

ToTabular automatically interprets the data as a column of "Boolean" type:

Take a collection of words or missing values:

ToTabular automatically interprets the data as a column of "String" type:

Take a collection of Quantity durations and missing values:

The values are stored using "Real64" magnitudes and a common unit of "Seconds":

For small enough Tabular objects, the original input data is cached by default and Normal recovers it:

Specify the magnitude type and unit to use:

The original input was not cached:

Quantities of different dimensions can be stored using the "InertExpression" type:

Trying to impose a specific "Quantity" type will result in missing values:

Take a collection of dates:

Create a Tabular object with automatic type detection:

Specify a type to convert to granularity "Day" when dates are extracted:

Take a collection of times:

Create a Tabular object with automatic type detection:

Specify a type to convert to granularity "Hour" when times are extracted:

Specify dates with a different calendar:

Take a collection of central geo positions for the countries of South America:

ToTabular automatically interprets the data as a column of "GeoPosition" type:

Take a collection of integer tuples:

Specify the type of the elements of the tuples:

Take a collection of tuples of objects of various types:

Create a Tabular object with automatic type detection:

Specify the types of the elements of the tuples:

Take a collection of regions:

Creating a Tabular object assigns type "Expression" to general Wolfram Language expressions:

Extract the component types of a TimeSeries object:

The association includes the type of the "Timestamp" column of the corresponding Tabular object:

Take a simple TimeSeries object:

The timestamp component has numeric or date type, while the value component can be of any type:

Get column types for all columns:

Extract column types for two columns:

Extract columns of Boolean type:

Extract columns of any integer type:

Extract columns that are either string or Boolean: