Question regarding executing auto-correlation

Hi all,

I have to stationarized my data using auto-correlation to be used in arima models. However, i have trouble getting the codes correct.

data = xlsread('20jun.xlsx');

pv = data; %144 readingfrom 20jun

t= [1:144]; %defined time points

plot(t,pv);

I have 144 data readings, resulting in 144 time point. Any advise will be appreciated. Thanks for the time.

[Mischa Kim]

What exactly is the problem with your code?

[chean hau]

Hi Mischa,

Thanks for taking the time to reply. The error i obtained is 'Vectors must be the same lengths'. Pardon my raw knowledge regarding matlab, but is it necessary to use an arima model to do acf? Thanks in advance.

Answer 1

[Image Analyst]

Before plot, put there lines and tell us what it says

size(t)  % Don't use a semicolon
size(pv)  % Don't use a semicolon
class(t)  % Don't use a semicolon
class(pv)  % Don't use a semicolon

And I don't see any autocorrelation going on, just reading data from a workbook and plotting it.

[chean hau]

Hi image analyst,

Thank you for taking time to read my post. Let me rephrase my poor post. I have these 144 readings(pv), and i need 144 time points to plot them. 1 time point to 1 pv reading.

size(t) is 1 144

size(pv) is 144 1

class(t,pv) are double.

I have actually tried using just one line, together with the above codes;

autocorr(pv)

which I doubt is correct, since the graph only showed 21 timepoints. Is there anyway I can rectify my codes?

[Image Analyst]

Try plot (t', pv) and try xcorr() with the 'coeffs' option. Look at the example in the help for xcorr for autocorrelation.

[chean hau]

Hi Image analyst,

Thanks for your input.

My codes are:.

x = xlsread('20jun.xlsx'); %144 readingfrom 20jun

t= [1:144]; %defined time points

plot(t,x);

figure;

[xc,lags] = xcorr(x,'coeff');

stem(lags,xc)

Is this acceptable? Sorry for my limited knowledge on matlab.

[Image Analyst]

I don't know. I don't have your data. Can't you tell if it works or not? Do you get the expected results?

[chean hau]

Hi Image analyst,

I think it works. Thank you so much for your advises. You have been a great help.