How to import csv-file with headlines (type text) to matrix? - only numbers are needed

Hello community,

I've got a csv-file with first row as headline (type text). But I want to write a script, so the import tool cannot help. I've tried 'importData', csvread (but it supports sheets with numbers only) and textscan. It would be the best when i could choose the range - I only need the numbers.

My csv-file has the following composition:

# | date | time | temperature 1 | temperature 2 | ..... | temperature 732

1 31-10-19 12:31:14 14,2342 15,1234 ....

2 31-10-19 12:31:15 16,4442 15,6283 ....

3 31-10-19 12:31:16 14,8552 15,8294 ....

I only need the temperature values, therefore i want to choose the range. Or is there anothers opportunity to import the numerical values?

Thanks for your help!

Michael

Answer 1

[Guillaume]

Both readtable and readmatrix will handle this file without issue. I would have thought that importdata would delegate to readtable so I'm surprise it didn't work for you.

readtable will label the columns according to the header.

[Michael Käufl]

Thank you for your answer. You're right, I get a table like the csv-table. But then I need a matrix with the number values. And table2array doesn't work like the way i want it to.

This error shows up:

error using table2array (line 37)

Unable to concatenate the table variables 'Var1' and 'Var2', because their types are double and cell.

Error in teeest (line 6)

matrix = table2array(test);

[Michael Käufl]

This is my data file (.csv). All I wanna have is the numerical values of each coloumn (but starting at coloumn 20) in a matrix. All values of one coloumn should be in one coloumn and the order of the coloumn must not change - that's important.

Could you give ma a 'route' how get to my target?

Cheers Michael

[Guillaume]

If all you want is a matrix from column 20, simply extract that from the table or matrix:

data = readtable('messung2.csv');
selecteddata = data{:, 20:1492};  %extract matrix from table. All selected columns must be numeric

%or
data = readmatrix('messung2.csv');  %non numeric columns in the file are NaN. 
selecteddata = data(:, 20:1492);

However, it seems a waste to discard the rest of the columns, in particular the date and time and to discard the column names, so you may want to actually give hints to matlab so it can parse your file properly:

opts = detectImportOptions('messung2.csv', 'VariableNamesLine', 1, 'ExtraColumnsRule', 'ignore');
data = readtable('messung2.csv', opts);  %Now it parses the file properly except for the date
data.Datum = datetime(data.Datum, 'InputFormat', 'dd,MM,yyyy');  %so parse the data
%and optionally merge date and time
data.Datum = data.Datum + data.Uhrzeit;

This table, which you could even convert to a timetable, would be very useful to perform aggregate calculation per day/month/year and make it easier to select columns based on name rather than assumed column number.

[Michael Käufl]

Perfect, thank you so much! :)