When trying to analyze multiple txt files from 1981-1983 using uigetfile, I get the error "unable to find or open temp_summary.01.1981". Check the path and filename"

I am trying to load multiple text files that has 4 columns (Day, High, Low, Average Temperature) and simultaneously compare each day's high, low, and average temperature of January 1st 1981 against January 1st 1982 through January 1st 1983 then repeat each step for January 2nd through December 31. This is to be in an array where I can find the max high, min low, low high, and high low temperatures and the corresponding date that it occurred. Loading the files is fine, but running it through the loop is an issue.

[file_list, path_n] = uigetfile('*.txt', 'Temperature Climatology', 'MultiSelect', 'on');

startYear = 1981;
endYear = 1983;

day = [];
highTemps = [];
lowTemps = [];
avgTemps = [];

for year = startYear:endYear
    for month = 1:12
        for day = 1:31
            filename = sprintf('temp_summary.%05d.%02d.%02d_%d.txt', month, year);
            data = readtable(filename);     <---------- this is where my error lies
            
            dayData = data(data(:,1) == day,:);
            
            highTemp = max(dayData(:,2));
            lowTemp = min(dayData(:,3));
            avgTemp = mean(dayData(:,4));
            
            highTemps(end+1) = highTemp;
            lowTemps(end+1) = lowTemp;
            avgTemps(end+1) = avgTemp;
        end
    end
end

highestHighTemp = max(highTemps);
lowestLowTemp = min(lowTemps);
lowestHighTemp = min(highTemps);
highestLowTemp = max(lowTemps);

monthlyHighAvg = [];
monthlyLowAvg = [];
monthlyAvgTemp = [];

for month = 1:12
    monthlyHighAvg(end+1) = mean(highTemps(month:12:end));
    monthlyLowAvg(end+1) = mean(lowTemps(month:12:end));
    monthlyAvgTemp(end+1) = mean(avgTemps(month:12:end));
end

moths = 1:12;
figure;
subplot(3,1,1);
plot(months,monthlyHighAvg, 'r');
title('Monthly Average High Temperature');
xlabel('Month');
ylabel('Temperature (°C)');

subplot(3,1,2);
plot(months,monthlyLowAvg, 'b');
title('Monthly Average Low Temperature');
xlabel('Month');
ylabel('Temperature (°C)');

subplot(3,1,3);
plot(months, monthlyAvgTemp, 'g');
title('Monthly Average Temperature');
xlabel('Month');
ylabel('Temperature (°C)');

[Stephen23]

Most likely the file does not exist (e.g. does not have that exact name) or is not located in that particular folder.

Your code has a number of bugs and quirks:

• Why do you call UIGETFILE but do not use its outputs? Perhaps you meant to use UIGETDIR together with FULLFILE (which would likely resolve any attempts to import files that are located in some other folder).
• Note that your code loops over dates that do not exist, e.g. February 30 & 31, April 31, June 31, etc.
• The SPRINTF format string implies that you expect four numeric values, but you call SPRINTF with only two scalar inputs. Lets see what happens when we try that:

month = 1;
year = 1998;
sprintf('temp_summary.%05d.%02d.%02d_%d.txt', month, year)

ans = 'temp_summary.00001.1998.'

Note how because you did not supply values for the 3rd and 4th fields the function simply returns at that point. Without the file extension! However most file have extensions, so this seems ... unlikely to be what you want.

Debug code. Start by looking at what your code is actually doing: what outputs do you get, for what inputs?

[Jonathon Klepatzki]

Thank you for the tip. I got ride of the uigetfile code and fixed the string (still working on mastering that) then ran it. Sadly, I have another error prior to fixing the dates that don't exist.

The new error starts two rows down (dayData = data(data(:,1) == day,:).

It states, a table row subscript must be a numeric array containing real positive integers, a logical array, a character vector, and etc... I'm assuming it's referring to data(:,1)??

[Walter Roberson]

Your data is the output of readtable() so it will be a table() object.

            dayData = data(data(:,1) == day,:);
            

When you use () indexing of a table object, the result is a table object. You are using == between a table object and a value. In a very recent version of MATLAB, that became a defined operation instead of an error, and it returns a table() object with the same variable names as the variable names in the table being compared to. So your data(EXPRESSION,:) has EXPRESSION as a table object... which is not a valid subscript.

Remember that () indexing of a table gives a table. Use {} indexing to get the result of a table --- or better yet, use dot indexing with a variable name, such as data(data.WeekDay == day,:)

[Jonathon Klepatzki]

Thank you for the help. You wasn't kidding about the number of bugs in this code.

Matlab is so fininicky. I change the code to data(data.Var1 == day,:);

And received the following error:

unable to perform assignment because value of type 'table' is not convertible to 'double'. Can you explain? Thank you again.

(What's interesting, it sees the table and the data is in it)

[Walter Roberson]

            highTemp = max(dayData.Var2);
            lowTemp = min(dayData.Var3);
            avgTemp = mean(dayData.Var4);
            

MATLAB is not being fininicky: MATLAB is being consistent.

Imagine you had a business ledger, and you extracted February 2022 information from it. Would the result be a (smaller) business ledger, or did the action of extracting some rows result in something that is not a business ledger but is something else (that is still structured with multiple columns of a variety of types) ? Now extract from that the columns having to do with date and postage expenses: is the resulting two-column thing with different types not also a business ledger? Now from that, do the same extraction process you used before to extract just the postage expense column: is the resulting one-column result not also a business ledger? If you used the same extraction process then the column will still have a name ("Postage Expenses") and associated values. If you want just the values, you would have to use a different extraction process -- one that left out the column name and just had the values.

Your code was doing the equivalent of using the extraction process that gets multiple columns data(data.Var1 == day,:) complete with headings, but was then expecting that if you used the same extraction process to extract a single column, that the result would contain just the values, where MATLAB says "when you use that extraction process, the result is always a table, and you need to use a different extraction process if you want just the content without the headers."

Could Mathworks have designed table objects such that if you used () notation to extract multiple columns that the result would be a table, but that if you used () notation to extract exactly one column, that the result would be the content without the header? Yes, of course it could have designed it that way.

But consider for example if you gave the task of extracting all of the table columns that show Overdraft, and you do not know ahead of time how many there are. Under the hypothetical design, if there happened to be multiple matching columns, then the result would be (hypothetically) be a table with the associated column names, and you would be able to tell what the overdraft pertained to. But if there happened to be exactly one matching column, then with that hypothetical design, the result would have to be just the content of the column without the header -- and you would not be able to tell what the data pertained to. You would not be able to predict whether the datatype returned would be table or content, so you would have to check class() of what was returned. And for the case where you wanted the headers left intact you would need to create an additional new extraction syntax for the case where you wanted to preserve the table structure even if it happened to be only a single column that was extracted.

It is a much cleaner design all around with the current syntax, where () of a table always results in a table, and you can optionally use {} indexing or dot indexing if you want to get at the contents of the columns.

[Jonathon Klepatzki]

That's awesome. Thank you.

Answer 1

[Walter Roberson]

[file_list, path_n] = uigetfile('*.txt', 'Temperature Climatology', 'MultiSelect', 'on');

You are ignoring both outputs of that

            filename = sprintf('temp_summary.%05d.%02d.%02d_%d.txt', month, year);
            data = readtable(filename);     <---------- this is where my error lies
            

and you are trying to read from a file in the current directory, whatever the current directory happens to be.

I suspect you should instead be usiing a uigetdir call, and that you should be using fullfile to put that directory together with the sprintf() result to get the name of the file to readtable()

By the way, remember to check whether the output of uigetdir() is character or not. If it is not character then the user used the "cancel" option of uigetdir(). For uigetfile() the situation is slightly more complicated: if the user cancels uigetfile() then the first output ( file_list in your code) will be numeric; if the user selects exactly one file, then the first output would be char; if the user selects more than one file, then the first output will be cell array of character vector. So after testing whether the output is numeric to check for cancel, use cellstr on the variable: if the variable is char then cellstr() will wrap it in a cell, and if the variable is already a cell then cellstr() will return it unchanged.