How do i convert a Python Pandas Dataframe to Matlab types?

I am calling a python function from Matlab code which returns a Pandas Dataframe. I would like to convert the content of the DataFrame to Matlab data types, but I can't find the correct way to do it. Thanks in advance for any help on this.

[per isakson]

See DataFrame by Nicholas

Answer 1

[Todd Leonhardt]

A simple and effective, but perhaps inelegant, solution is to first save the Pandas DataFrame as a CSV file and then read that data into MATLAB as a table datatype (as long as you have MATLAB R2013b or newer).

You can save a Pandas DataFrame as a CSV file with the DataFrame.to_csv() function: http://pandas.pydata.org/pandas-docs/version/0.18.1/generated/pandas.DataFrame.to_csv.html

You can read a CSV file into MATLAB as a table using the readtable() function: http://www.mathworks.com/help/matlab/ref/readtable.html

[Danilo Botero Lopez]

It worked!

Answer 2

[Todd Leonhardt]

Another possibility is to convert the Pandas DataFrame to a Python dictionary using the DataFrame.to_dict() method: http://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.to_dict.html

Once you have the data as a Python Dictionary, you should be able to a use that in MATLAB as shown here: http://www.mathworks.com/help/matlab/matlab_external/use-python-dict-type-in-matlab.html

[Danilo Botero Lopez]

Thank you!

Answer 3

[Jeffrey Daniels]

I solved this problem by querying the Pandas dataframe column data types and inserting them as the last row of the dataframe, convert the df to string and pass to Matlab. Then once in Matlab I have a method that reads the string into a Matlab table and applies the data type specified in the last row of the CSV to each column of the table. It's not elegant, but it is fairly bulletproof. This method works completely in memory space and does not require writing to hard drive.

[Dev-iL]

Would it be possible for you to post a code sample? There are several moving parts in this solution...

Answer 4

[Artem Lensky]

Check this submission https://au.mathworks.com/matlabcentral/fileexchange/111770-pandastomatlab

It implements two funcions:

• df2t - that converts Pandas DataFrame to Matlab Table
• t2df - that converts Matlab Table to Pandas DataFrame.

The examples are shown in test.mlx.

Answer 5

[Bryan Yu]

HDF5 is native for both pandas and MATLAB. Could be a good compromise for speed and type safety.

Answer 6

[Francesco Giaretta]

Starting from MATLAB R2024a, is it possible to convert a Pandas DataFrame directly into a Matlab table:

https://it.mathworks.com/help/releases/R2024a/matlab/release-notes.html?rntext=&startrelease=R2023b&endrelease=R2024a&groupby=release&sortby=descending&searchHighlight=#mw_7ad2f7e1-43a2-4121-8b53-88a9a0660554