How to avoid variable name warnings when using filenames as variable names

Hi,

I have almost got the orange errors at the side of my script down to zero but have one error that keeps coming up: "Warning: Variable names were modified to make them valid MATLAB identifiers. The original names are saved in the VariableDescriptions property. "

Using isvarname I have traced the error to the excel filenames my script reads in and then processes (I need to keep the filenames as is). There was a decimal point which I can exclude, but I cannot exclude the '.xlsx', or do not know how to. Below is a simple example, can anyone show how this file could be read in with a variable title 'stuff1_2_3'?

Best regards

Steve

close all
clear all
clc

s='stuff1_2_3.xlsx'
isvarname(s)

[Cedric]

I think that your approach is wrong if you need a variable name to match a file name. What are you trying to achieve?

[Stephen Devlin]

The filename contains the testing information particular to that trial and is used as a string to annotate the graphs, otherwise the graphs would mean nothing to those looking at them. The script works it just comes up with some orange errors.

[Cedric]

You are using tables, so I understand that you have limited control on the import. I am not using tables (I don't think that they are mature enough yet) so I don't know how to manage this properly using them.

Generally though, we avoid using inputs as field names, precisely because they are likely not to be compatible with field name requirements. Instead, we use a cell array of e.g. column headers that matches e.g. the number of columns of a numeric array or a cell array, and we use the headers just for display (e.g. plot titles). when we are dealing with advanced users and we want to provide e.g. a struct with field names that correspond to column headers, we "normalize" the headers so they become compatible with field name requirements. There are MATLAB functions for this, but it is easy to produce using a few operations on strings.

Answer 1

[Jeremy Hughes]

The warning you're seeing comes from READTABLE; it's probably not related to the names of your files (unless those are being stored in the XLSX file).

Find the call to readtable, and then add the parameter 'ReadVariableNames',false ... e.g.

t = readtable(filename,'ReadVariableNames',false);

This will make the warning go away, but I don't know if that's the behavior you want--the variable names of the table will look like 'Var1', 'Var2', ... etc..

Hope this helps, Jeremy

[Stephen Devlin]

Hi Jeremy,

The portion of code reading in the excel files is:

Folder_used = dir('/Users/imagexpertinc/Desktop/stuff/*.xlsx');
files = Folder_used;

for k=1:numel(files)

data=readtable(files(k).name);

do stuff
end

[Stephen Devlin]

I also have column headers within the files that are lost if I set the ReadVariableNames to false.

[Jan]

You can use xlsread to keep the header line. But there is no automatic conversion of these names to Matlab variables, if the names do not allow this. You have to store the names separately.

[Jeremy Hughes]

It's the column headers in your file being converted to table variable names that is causing the warning. (Variable is the terminology used by table to name the "columns" since each variable could itself have multiple columns) As the warning says, you can get the original data from the table's VariableDescriptions property. (data.Properties.VariableDescriptions) Should you need it.

The column headers in the file probably have spaces, leading numbers or other characters which are not allowed in variable names. The warning here is just an alert that the data has been modified so table can use those names. A warning isn't the same as an error; There's nothing going wrong, per se. There's just a condition you should be aware of.

[Stephen Devlin]

You were right, one of the headers had the units as (m/s) when the equipment was supposed to represent that as _ms, fixed that and no more orange errors. Thank you Jeremy!

[Maximilian Prilmueller]

You might want to try it with the 'VariableNamingRule', 'preserve' option. This will probably mess up your column names a bit, but will get rid of the warnings. And a messed up name is still better than none, I guess.

[Zeinab Alaraji]

Thanks , Thats worked correvctly , PreservedNamedVaraible to be true

Answer 2

[Jan]

The name of an Excel file need not and should not be a valid Matlab name.

s    = 'stuff1_2_3.xlsx'
Data = xlsread(s);

Now Data contains the contents of the Excel file, which name is stored in the variable s. I cannot imagine why you want to use the file name as a variable.

Please show us the code, which causes the warning.

[Stephen Devlin]

Hi Jan,

The filename is the only way the persons carrying out the tests can associate operational data with the results data, so a typical files name might be something like P101_10V_10KHZ_J231_SF0.98

The SF0.98 also caused the same warning, so I took the decimal point out, this halved the numbers of warnings. The script is to process a batch of excel files in a loop, it works fine but the warning is just annoying.

[Jan]

If the filename is important, store it - but not in name of the variable. It is prone to errors and inefficient to use dynamic names for variables and the important information is simply to hard to access. Prefer:

FileName         = 'stuff1_2_3.xlsx'
Measurement.Data = xlsread(s);
Measurement.Name = FileName;

Like in the real world: You do not include the weight, driving license ID and birthday in your name also.

Answer 3

[john greene]

tables have a header row. the readtable command reads in the table and the header row text is assigned to variables in matlab. if the text entries in your header row are not valid matlab names, matlab automatically gives the invalid ones new names that are valid as a matlab variables. for example, if one of your header row labels has a question mark (?) in it, it will cause matlab to change the variable name and issue the warning you're asking about.

Answer 4

[Weifu]

It probably because the item title does not compliant with the variable rule and you need to modify the names. i.e "item 1" is not valid and it need to change to item1 or item_1.