Cell array of dates in string format to date array

10 views (last 30 days)

I've pulled some data from a spreadsheet and created a new cell array from the first column, which is dates in the format 'dd/MM/yyyy' (data attached) and then I've tried to convert those to dates:

[~, ~, raw] = xlsread('file.xlsx', 'sheet');
date = {rawFull{2:end,1}}';
date2 = datetime(date, 'InputFormat', 'dd/MM/yyyy');

But I get the following error:

Error using datetime (line 629)
Input data must be a numeric or a date/time string or a cell array or char matrix containing date/time character vectors.

I've had a look using a smaller dummy array that works correctly:

dummyOld = {'01/10/2017'; '01/10/2017'};
dummyNew = datetime(dummyOld, 'InputFormat', 'dd/MM/yyyy');

I assume the difference between working and not working is something to do with the quotation marks shown below, but I'm not sure how to overcome that? Any help (with an explanation of what I'm missing) is much appreciated!

  3 Comments
Lauren Dransfield
Lauren Dransfield on 1 Oct 2018
Edited: Lauren Dransfield on 1 Oct 2018
Apologies, data uploaded.
Thanks for your suggestion, but I'd prefer to avoid using tables if possible.
jonas
jonas on 1 Oct 2018
Edited: jonas on 1 Oct 2018
No problem, if you want to avoid tables that's fine. One of the advantages with tables is that you can mix different classes, so you do not have to work with "raw data". You can easily transfer the data from a table to a struct or cell array post-import if you prefer to work with such variables.

Sign in to comment.

Accepted Answer

jonas
jonas on 1 Oct 2018
Edited: jonas on 1 Oct 2018
The problem is that you have a number of NaNs at the end of your cell array. Due to those NaNs, you can not concatenate the data in a single array (you cannot mix strings and doubles). It's best to avoid this issue altogether by importing the data properly. For example, import only the relevant cells when calling xlsread or, better yet, use readtable.
Upload the xlsfile if you need more help!
If you really want to proceed with your previous approach, then find and delete the cells containing NaNs prior to calling datetime
data=load('matlabDateParseError.mat')
d=data.date;
id=cellfun(@(x)sum(isnan(x)),d)
d(logical(id))=[];
datetime(d)
don't name your variable date! It's a built in function.
  2 Comments
Lauren Dransfield
Lauren Dransfield on 1 Oct 2018
Thanks, Jonas
I had something removing the NaNs afterwards so a bit of reorganising worked a treat

Sign in to comment.

More Answers (1)

Peter Perkins
Peter Perkins on 1 Oct 2018
Lauren, is there a reason you want to avoid using tables? You really will be much happier using readtable, especially in recent versions, even if just to do the import.

Categories

Find more on Data Type Identification in Help Center and File Exchange

Products


Release

R2017a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!