Convert Between MATLAB tables and Pandas DataFrames

MATLAB^® Compiler SDK™ supports conversion between tables and Pandas DataFrames. This process also converts data inside the DataFrame or table implicitly. DataFrames can also be converted to timetables and vice versa when the index of the DataFrame is a datetime object. To convert between other types of data in MATLAB and Python^®, see Pass Data Between MATLAB and Python. For an example of conversion between MATLAB tables and Pandas DataFrames, see Use MATLAB Compiler SDK to Work with Pandas DataFrames.

Note

MATLAB Compiler SDK supports only one level of conversion with DataFrames. It does not support recursive nesting. For example, you can create a DataFrame df1 containing another DataFrame df2 and convert it to a table. However, you cannot convert df1 if df2 contains another DataFrame df3.

Convert DataFrames to Tables

The list below includes all data types that, when contained within a Pandas DataFrame, are supported for conversion to a MATLAB table.

Python Data Type	Numpy Data Type	Pandas Data Type	Resulting MATLAB Data Type
	`numpy.float64`	`pandas.Float64`	`double`
	`numpy.float32`	`pandas.Float32`	`single`
	`numpy.uint8`	`pandas.UInt8`	`uint8`
	`numpy.uint16`	`pandas.UInt16`	`uint16`
	`numpy.uint32`	`pandas.UInt32`	`uint32`
	`numpy.uint64`	`pandas.UInt64`	`uint64`
	`numpy.int8`	`pandas.Int8`	`int8`
	`numpy.int16`	`pandas.Int16`	`int16`
	`numpy.int32`	`pandas.Int32`	`int32`
	`numpy.int64`	`pandas.Int64`	`int64`
	`numpy.complex64`	`numpy.complex128`	complex (`single`, `double`)
`bool`			`logical`
		`pandas.Categorical`	`categorical`
	`numpy.datetime64`	`pandas.Timestamp`	`datetime`
	`numpy.timedelta64`	`pandas.Timedelta`	`duration`
	`object`	`object`	`string` (if the elements of the array are strings, otherwise not supported)
`str`		`pandas.StringDtype`	`string`
Sequence (e.g. `tuple` or `list`)			`cell`
`dict`			`struct`
		`pandas.DataFrame`	`table`
		`pandas.DataFrame` with `datetime` index	`timetable`

Convert Tables to DataFrames

The list below includes all data types that, when contained within a MATLAB table, are supported for conversion to a Pandas DataFrame.

MATLAB Data type	Resulting Numpy or Pandas Data Type
`double`	`numpy.float64`
`single`	`numpy.float32`
`uint8`	`numpy.uint8`
`uint16`	`numpy.uint16`
`uint32`	`numpy.uint32`
`uint64`	`numpy.uint64`
`int8`	`numpy.int8`
`int16`	`numpy.int16`
`int32`	`numpy.int32`
`int64`	`numpy.int64`
complex (`single`)	`numpy.complex64`,
complex(`double`)	`numpy.complex128`
`logical`	`bool`
`categorical`	`pandas.Categorical`
`datetime`	`numpy.datetime64`
`duration`	`numpy.timedelta64`
`string`	`string`, but the type for the whole Series will be `object`
`cellstr`	`string`, but the type for the whole Series will be `object`
`cell`	Resulting type varies depending on the elements of the `cell` array, but the type for the whole Series will be `object`
`struct`	`dict`, but the type for the whole Series will be `object`
`dictionary`	`matlab.dictionary`, but the type for the whole Series will be `object`
`table`	`pandas.DataFrame`
`timetable`	`pandas.DataFrame` with datetime index