In MATLAB (cumulative distribution function), how can I find the corresponding data point (Y) for any chosen cumulative probability?
18 visualizzazioni (ultimi 30 giorni)
Mostra commenti meno recenti
Aaron
il 29 Mag 2013
Commentato: Image Analyst
il 27 Lug 2018
In a CDF (using MATLAB,) how can I find the corresponding data value (X) for any chosen cumulative distribution (Y)? Please refer to the pasted code and image. Instead of "eye-balling" the plot, how can I find the data point (X) that corresponds to the cumulative probability value of 0.2 or even 0.466, etc.? Please advise. Thank you.
X = randn(1,500);
u = mean(X);
s = std(X);
pd = makedist('Normal','mu',u,'sigma',s);
x = min(X):.1:max(X);
cdf_normal = cdf(pd,x);
plot(x,cdf_normal,'LineWidth',4)
0 Commenti
Risposta accettata
Teja Muppirala
il 29 Mag 2013
Alternatively, you can use ICDF directly on the probability distribution object
icdf(pd,[0.2 0.466])
0 Commenti
Più risposte (2)
Image Analyst
il 29 Mag 2013
Modificato: Image Analyst
il 29 Mag 2013
If "the_Y_Value" is the y value you're looking at, and xValue is the x location where y first exceeds "the_Y_Value", then:
xIndex = find(cdf_normal >= the_Y_Value, 1, 'first');
xValue = x(xIndex);
0 Commenti
Brett Hoover
il 26 Lug 2018
Kinda surprised that nobody suggested 1-D interpolation:
target_value = interp1(cdf_probs,cdf_values,target_prob)
1 Commento
Image Analyst
il 27 Lug 2018
Because when Aaron said "how can I find the data point..." I assumed he wanted a data point that was in his existing set of data. Your code would give a point not in the data. It's probably a more accurate estimate but wouldn't be one of his data points -- it would be between two of them. However, maybe that's what he really meant.
Vedere anche
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!