Plot large CSV files with timestamp

I have a large csv file and try to load it into MATLAB and plot it with timestamp. My release is 2016b and i have 8gb of RAM.

My code starts like this

ds = tabularTextDatastore('2018-11-18.csv','TreatAsMissing','NA')

data = preview(ds);
whos data

data(:,1:7) % Look at first 7 variables

testdata = read(ds)

[Jan]

What is your question?

[Ancalagon8]

I need help loading and plotting my large csv file with timestamp.

Answer 1

[Walter Roberson]

strcat the first two columns together with a blank between . datetime the result with an InputFormat parameter , which is required for this purpose because datetime cannot figure out fractions of a second by itself .

Now that you have timestamps and values you can prepare to plot . I think your release permitted using datetime as the X axis .

[Ancalagon8]

As I have daily files I only need the hour and not the date, so starting point will be the second collumn. Can you help me with the code?

Thanks in advance!

[Walter Roberson]

That is more work. Time of day without an attached date is handled by duration objects. In your R2016b release, duration() did not support input as time strings, so to construct the objects it would be necessary to parse the character strings into arrays of numeric values that can then be passed to duration() . It is easier to

dt = datetime( strcat(testdata{:,1}, {' '}, testdata{:,2}), 'inputformat', 'yyyy-MM-dd HH:mm:ss.SSSSSS');

You can always convert those to duration afterwards:

dur = dt - dateshift(dt(1), 'start', 'day');

After which you can plot, such as

plot(dur, testdata{:,3:4})

[Ancalagon8]

and testdata is my file?

[Walter Roberson]

Like you posted,

testdata = read(ds);