Load function not properly loading a numeric ASCII file

I am trying to load and ASCII file that is 24X48. The load function will not load the load the actual values in the work space. It will load zeros. The "readtable" will work without a problem but it puts in format has to be converted into an array and can only be done after individual values are pulled out, put into a 1X1 table then converted into an array. I am using R2018a at this time. I never had this problem with my old version R2008b.

Is there a way to use the load function on a ASCII file that is in the form of a table that will not cause this issue?

Ideally, I need a function that can load in an ASCII file into the workspace as an array quickly without jumping through hoops and allow the data to be either a character or an integer or number?

[Matthew Benedict]

You may want to take a look at this function that can handle mixed types in a delimited file: https://www.mathworks.com/matlabcentral/fileexchange/10946-readtext

[Walter Roberson]

Please attach an example of the file.

If you have a table, T, that consists entirely of numeric values, then

T{:,:}

would be a numeric array of the values.

[Eric Metzger]

Here is an example of the data and how it is formatted.

[Walter Roberson]

Looks to me as if it might be tab delimited. I do not see any reason for load() to load it as zeros, though. It would be better if you could attach an extract rather than just an image of a portion of it.

[Eric Metzger]

Here is the full file attached. This file is tab delimited. I have used this type of file many times over with R2008b but some reason, the new version (R2018a) will not load this correctly.

[Walter Roberson]

Try giving the command

format long g

and then displaying the data again.

[Eric Metzger]

After some more research, the .txt files in question originate from a Unix/Lnix work station. It may have somthing to do with how the file "new line" command works between Unix and PC.

I have also looked at and tested the 'readtext' function mentioned in the first comment. This will read the files but the function puts it at a cell array not a standard array or .mat file. This is workable but I have to pull each cell out separately and then do a number of steps to get them into a standard .mat format so I can run my calculations on them. This is not ideal as it requires a number of different steps to get it to work. Ideally, I would like a function that will just read any text in as a standard array (.mat) file.

[Star Strider]

The load function imported it correctly when I tried it. (See my Answer.)

[Walter Roberson]

I just tried it on R2018a on Windows 10. load() worked fine for me. I did have to use

format long g

to be able to see the full range of values.

Answer 1

[Star Strider]

Your file is (24 x 47), and load reads it correctly and completely:

D = load('20180919N.txt');

I copied your entire file to my workspace, as well as imported it with load, then did a logical test for equality between the two matrices. The two matrices are the same.

[Eric Metzger]

Tried many times, it still loads the file as zeros. Here is the file in question and a screen capture of what I get when I load it. I am only having this problem with txt files originating from a Unix/Lnix based system to a PC based system. Any txt file that originates on a PC, I don't have this problem. This makes no sense why this would be happening.

[Star Strider]

They are not all zeros.

In the upper left corner:

x = 

   1E+09 *

and ‘x(:,3:4)’ are therefore 2.0181E+9. The other values are less than 1E+5, so they simply display as zeros.

It appears to be loading them correctly. If you wan t to see all the details, before you read the file, specify:

format long g

When I do this:

x = load('20180929n.txt');
format long g
x

I get this:

x =

    Columns 1 through 4

                       37.66                    -97.44                2018092907                2018100207
                     39.0294                    -98.13                2018092907                2018100207
                     38.3614                   -98.828                2018092907                2018100207
                     38.2936                  -97.9113                2018092907                2018100207
                     37.6454                   -98.124                2018092907                2018100207
                      38.078                  -96.8877                2018092907                2018100207
     ...

So load is importing your data. Your current format setting is simply not displaying it all. There is nothing wrong anywhere, as I see it.

[Eric Metzger]

Hmmm....Knew it had to be something simple. Is there a setting in MatLab to display at "long" by default. I have looked at I can't seem to get it to do that. WHile this may be simple, this does cause a problem when working with the data. So I need to some how prevent Matlab from assuming I want it in this format. I prefer to see it not in Sci Notation unless I am dealing with numbers with more than 12 digits, which is not common for me.

[Eric Metzger]

Ok. Thanks for the help. Leave it to me to have an issue that is an easy fix. LOL

[Star Strider]

You can easily switch between format options. I usually have mine set to:

format short eng

so when I read your file and displayed it, I wrote:

x = load('20180929n.txt');
format long g
x
format short eng

That displayed it as I wanted, so I could post the result, and then reset it to my desired option. See the documentation on format (link) for all the available options.

To set the default format:

[Walter Roberson]

... format long g is what I had suggested a week ago.

[Eric Metzger]

Wanted to thank all that contributed to this question. While I have used MatLab in the past, I am not just getting back into it and some of the simple issues like format I am having to relearn. Would have posted this sooner but I have been dealing with a number of issues. This is the first chance I have had to get back to this. This did work and it is now a standard in my scripts.

[Star Strider]

Our pleasure.

MATLAB has changed over the years, and even between R2018a and 2018b. See the Release Notes (link) for a release-by-release synopsis. It now also has a search function.