Featured post
matlab - Calculating confidence intervals for a non-normal distribution -
first, should specify knowledge of statistics limited, please forgive me if question seems trivial or perhaps doesn't make sense.
i have data doesn't appear distributed. typically, when plot confidence intervals, use mean +- 2 standard deviations, don't think acceptible non-uniform distribution. sample size set 1000 samples, seem enough determine if normal distribution or not.
i use matlab processing, there functions in matlab make easy calculate confidence intervals (say 95%)?
i know there 'quantile' , 'prctile' functions, i'm not sure if that's need use. function 'mle' returns confidence intervals distributed data, although can supply own pdf.
could use ksdensity create pdf data, feed pdf mle function give me confidence intervals?
also, how go determining if data distributed. mean can tell looking @ histogram or pdf ksdensity, there way quantitatively measure it?
thanks!
are sure need confidence intervals or 90% range of random data?
if need latter, suggest use prctile(). example, if have vector holding independent identically distributed samples of random variables, can useful information running
y = prcntile(x, [5 50 95])
this return in [y(1), y(3)] range 90% of samples occur. , in y(2) median of sample.
try following example (using distributed variable):
t = 0:99; tt = repmat(t, 1000, 1); x = randn(1000, 100) .* tt + tt; % simple gaussian model varying mean , variance y = prctile(x, [5 50 95]); plot(t, y); legend('5%','50%','95%')
- Get link
- X
- Other Apps
Comments
Post a Comment