Memory issues in using xlsread and dataset('XLSFile'...)

I have a huge file in XLSX format. While xlsread runs without memory issues, the command A = dataset(XLSFile, filename) throws the error "Error: Not enough storage is available to complete". Why would this happen ? Any broad guidelines to keep in mind while handling large data will be very helpful.

Thank You

[Peter Perkins]

Vandhana, it would help to know more specifically what your data are, and the error trace that gets spit out to the command window, if there is one.

Answer 1

[Walter Roberson]

I speculate that dataset() reads the data and then converts it in to a different organization, requiring intermediate storage equivalent to the size of the data.

I do not know this to be true, as I do not have that toolbox, but from a code point of view it is more economical to only write the xlsx reading code once and convert the output, rather than writing code to read xlsx for every routine that is going to convert the data, just to prevent potential problems with intermediate storage being exceeded.

Answer 2

[Matt Tearle]

Do you have a simple way to write the file out in a different format (eg text, such as CSV)? What are the columns of your file? (ie how many and what type)

If you can work with text, you might be able to batch process and/or use non-double types to reduce the memory requirements.

Answer 3

[Nick Haddad]

This issue is a known bug in MATLAB and has been addressed in the following bug report:

• xlsread might return an error for large spreadsheets

The bug report has a workaround which you can install for MATLAB R2013a through R2014b.