Averaging data

Hello,

I'm having difficult time trying to average my data. I have data which look like this

 time1 = [0.14 0.99 1.67 2.45 3.10 3.70 4.24 5.00]
 res1  = [4 6 2 9 1 0 4 7]

 time2 = [0.11 0.69 1.00 1.45 1.66 2.04 2.24 2.99 3.11 3.43 4.25  
          4.55 5.00]
 res2  = [8 1 5 3 7 1 3  10 9 5  3 2 1]

 time3 = [0.09 0.33 1.13 1.44 2.10 2.70 3.24 4.00 4.80  5.00]
 res3  = [0 3 4 7 2 6 3 4 9 8]

The period of time is always 5 min but there is different number of responses given during this time. These responses showing only a change at particular time and the value between time intervals is constant. I'd like to plot average of the data using a stairs function. Any help highly appreciated.

Answer 1

[Oleg Komarov]

time1 = [0.14 0.99 1.67 2.45 3.10 3.70 4.24 5.00];
res1  = [4 6 2 9 1 0 4 7];

axis([0 5 0 10])
hold on
% Plot stairs
stairs(time1,res1)
% Plot average
avg = sum(diff([0 time1])/time1(end) .* res1);
line([0,time1(end)],[avg avg],'Color','r')

One clarification, is the first element of res1 - 4 - valid from 0 to 0.14 or as it is it's ok?

EDIT

% unique time vector
 untime = unique([time1 time2 time3]);

% Idx 
[trash,idx1] = histc(untime,[0 time1(1:end-1) inf]);
[trash,idx2] = histc(untime,[0 time2(1:end-1) inf]);
[trash,idx3] = histc(untime,[0 time3(1:end-1) inf]);

rep = [res1(idx1); res2(idx2); res3(idx3)];
avg = mean(rep);

hold on
% Plot all the lines
stairs(untime,rep.')
% Plot the average in black
stairs([0 untime], [9 avg],'Color','k','Linew',2)

[Marcel]

Thanks, but stairs function is not a problem here.

I'd like to plot an average of all data, not only time1 vs.res1.

[Oleg Komarov]

Can you be more specific? I added a line which is the average of res1, is it still not what you want?

[Marcel]

ok, I'll try to describe it more in details.

As I wrote above my data is not only time1 and res1. I have three responses from three diffrent subjects(res1, res2, res3) within 5min time intervals(time1, time2, time3). I'd like to get an average of these three responses. The responses are some sort of scores which are given at specific time (for example 4 is given at 0.14min, 6 at 0.99min - between 0.14 and 0.99min the score is 4 and the change to 6 happens at 0.99). In other words, if you plot these three sets of data (time1, res1; time2, res2; time3, res3) in the same figure, then what I'm looking for is the average of these three plots.

Hope it's more clear now.

[Marcel]

I forgot to put 0min in the time vectors, at the start(0 min) the value is always 9.

[Oleg Komarov]

See my edit.

[Fangjun Jiang]

Brilliant, Oleg! I was thinking about interp1() but it doesn't allow duplicated x values. I assume untime is something like untime=0:0.01:5. You might want to add that to avoid another question from the OP.

Also, consider adding data to make res=9 when time=0 (indicated by the OP) for all three series and modify your histc() lines too.

[Fangjun Jiang]

or untime=unique([time1,time2,time3]) should work and is better.

[Oleg Komarov]

forgot to post it...it was exactly that!

Answer 2

[Marcel]

Thanks very much guys. This is exactly what I was looking for. I realy appreciate your help.

[Marcel]

I have one more question. How to average the data within .5 min time intervals. I mean, to get one average point for every .5min across all responses. Having these average points I'd like to plot it using stairs function with error bars for each point.

Thanks in advance

[Oleg Komarov]

Then create:

untime = 0:0.5:5

instead of untime = unique(...);

[Marcel]

Thanks Oleg, you are very helpfull.

What is the best option to plot error bars in my case? I was trying to use errorbar function but did not work very well for me...

[Oleg Komarov]

What behaviour did you observe with error bar that didn't satisfy you?