MATLAB: How to convert a dataset array into a cell array in Statistics Toolbox 6.2 (R2008a)

arraycelldatasetStatistics and Machine Learning Toolbox

I am using dataset arrays to manipulate data that I pull from a database. After manipulation, I want to convert it back into a cell array and put it back into the database.

Best Answer

Beginning with MATLAB R2012b, you can use the "dataset2cell" command. For previous product releases, read below for any possible workarounds:

The ability to convert a dataset array directly into a cell array is not available in the Statistics Toolbox.

As a workaround, write a conversion function that accesses the data within a dataset array and puts the elements into a cell array. An example of this is attached as ds2cell.m.

Please note that this function assumes your data set has both variable and observation names. You may have to modify this function to fit your particular application.

Related Solutions

MATLAB: Is it possible to convert a cell array with multiple datatypes to a dataset array in Statistics Toolbox 7.1 (R2009a)

Use the CELL2DATASET command in MATLAB 8.0 (R2012b) to achieve this.

For previous releases of MATLAB, read below for any possible workarounds:

The ability to convert a cell array to a dataset array is not available in the Statistics Toolbox 7.1 (R2009a).

As a workaround, the attached MATLAB file 'cell2ds.m' can perform this conversion. It takes as input an MxN cell array, and outputs a dataset array of similar dimensions. The second optional input argument is a cell array of strings, containing the names of the columns. If this parameter is omitted, then the function will auto-assign column names in the output dataset array as 'col1', 'col2', etc.

If "test" is the name of cell array, and if it contains 7 columns, then 'cell2ds' is called as follows:

ds = cell2ds(test)   % leave out column names

Optionally, you can choose to include the column names:

ds = cell2ds(test, strings {'A','B','C','D','E','F','G'} )

The 7 strings {'A','B','C','D','E','F','G'} represent the names of each column.

MATLAB: Does the model initialize slowly in Simulink 6.2 (R14SP2)

This bug has been fixed in Release 2006a (R2006a). For previous product releases, read below for any possible workarounds:

There is a bug in Simulink 6.2 (R14SP2) in the way that it handles port dimensions, especially for large models. Sometimes the port dimensions can take a long time to resolve leading to slow model initialization.

To work around this issue, you may wish to place Signal Specification blocks in your model, away from where the signal dimensions are already specified. Try using these blocks especially after the signal has passed through multiple MUX and DEMUX blocks.

Best Answer

Related Solutions

MATLAB: Is it possible to convert a cell array with multiple datatypes to a dataset array in Statistics Toolbox 7.1 (R2009a)

MATLAB: Does the model initialize slowly in Simulink 6.2 (R14SP2)

Related Question