Plot time series with irregular intervals and interpolate at regular intervals

Hello all,

I have a matrix which contains data in this format: [ Date Time(HHMMSS) Price ]

2  70052  40.175
2  70214  40
2  70249  40
2  70347  40
2  70400  40

I want to be able to plot this as a time series as is, and then do linear interpolation for some interval (e.g. 15 mins) to get a reduced matrix. How may I be able to do this? I tried plotting the 3rd column but it becomes equally spaced 1 second intervals.

Thank you,

Answer 1

[Jonathan Epperl]

You plot minutes vs. price by running (Assuming your data is in the variable M)

plot(M(:,2),M(:,3));

With regards to the interpolation you need to be a little more specific. You fit the best (in the least-squares sense) straight line to your data by running

r = [M(:,2) ones(size(M,1),1)]\M(:,3)
% Check v your Data
plot(M(:,2),M(:,3),'kx',M(:,2),r(1)*M(:,2)+r(2),'r');
legend('Data', 'Linear Interpolation')

so that r(1) is the slope, r(2) the intercept of the line.

[Louise]

Hello, I've tried simply plotting the 2nd column against the 3rd column but the x axis includes the points when the minutes of the time (HHMMSS) is greater than 60 minutes, eg. 76100 (7:61:00) or 87100 (8:71:00) and I don't know how to get rid of it.

[Jonathan Epperl]

Could you elaborate on that? The middle column is supposed to be HHMMSS, but there are entries where the 2nd to 3rd digits are >59? How could that happen? Do you want to simply discard those points, or do you want to convert those so they make sense?

[Matt Fig]

Jonathan, read the time from the last digit. The leading 0 is absent....

70052 = 07:00:52

[Jonathan Epperl]

Matt, like I was saying, 2nd to 3rd digits are >59? Still, how is that possible if those are actually hours, minutes and seconds without the separators?

[Matt Fig]

Oh I missed the first comment. I thought you were seeing that in the OP.

[Louise]

So this is what my plot looks like:

I do not have any points where the minutes or seconds are > 60, but the graph includes those times in the axis and I'm not sure how I can get rid of it.

[Walter Roberson]

seconds = fix(M(:,2)/10000) * 60 * 60 + fix(mod(M(:,2),10000) / 100) * 60 + mod(M(:,2),100);
plot(seconds, M(:,3))

[Louise]

With your code, the x-axis still shows minutes/seconds >60. I managed to convert all the times into the datenum format. I don't know if there is something to be done then.

EDIT: Okay nevermind. I managed to use datetick() to change all the dates in datenum() format, so it displays the time correctly. However, I am not sure if using interp1() on the dates in datenum format will mess things up.

[Walter Roberson]

The code I gave converts everything to seconds, not to minutes/seconds. You need continuous ranges in order to interpolate. The serial date format such as is returned by datenum is fine for interpolation.