Question

I am working with a meteorological dataset and need to extract one column from many csv files and compile the results into a new file. I have it working through one month but the script gets stuck when it encounters a shorter month (Subscripted assignment dimension
mismatch). Makes sense, however, I want it to simply keep the NaN values that are originally present in the placeholder matrix D.

Here is the problem part of the script,

%Convert dates to matlab date numbers and get number of rows
Date = datenum(Date{1, 1}, 'dd-mm-yyyy');
T = size(Date, 1);    

%# Preallocate a matrix to hold all the data, and add the date column
D = [Date, NaN(T, NumFile)];

%# Loop over the csv files, get the eleventh column and add it to the data matrix
for k = 1:NumFile
FileList = dir('*.csv');
NumFile = size(FileList,1);
    filename = FileList(k).name;
    disp(filename);
    %# Get the current file name
    CurFilePath = filename;

    %# Open the current file for reading and scan in the second column using numerical format
    fid1 = fopen(CurFilePath, 'r');
    CurData = textscan(fid1, '%*s %*s %*s %*s %*s %*s %*s %*s %f %*[^\n]', 'Delimiter', ',"', 'HeaderLines', 17, 'MultipleDelimsAsOne',true);
    fclose(fid1);

    %Add the current data to the cell array
    D(:, k+1) = CurData{1, 1};
end

So, how can the shorter months be forced to the size of 31 days months to fit the placeholder matrix D.

Answer 1

When you are assigning D with the colon operator in one dimension, Matlab has to assume that you are assigning all elements in the row. To fix it, simply exchange the colon with a 1:numberOfDaysInMonth.That way Matlab will only assign the number of values you specify and leave the rest unchanged, Nan in this case.

numberOfDaysInMonth you can calculate as size(CurData{1, 1},1)

In all, exchange that next-to-last line in your script with:

%Add the current data to the cell array
numberOfDaysInMonth = size(CurData{1, 1},1);
D(1:numberOfDaysInMonth, k+1) = CurData{1, 1};