How to calculate the 95% (area under the curve) of Kernel probability density curve?

Hi everyone,

I require to plot the kernel density curve and mark its 5% and 95% values on the plot. I attempted but not working (my script is attached) May someone suggest to me how can I fix this?

raw=readmatrix('5_95.csv');

a=raw(:,4);
figure(1)
pdSix = fitdist(a,'Kernel','Width',0.5);
x = -5:0.01:5;
ySix = pdf(pdSix,x);
plot(x,ySix,'k-','LineWidth',2)

The espected output should be like this.

Answer 1

[Paul]

The function icdf can find the values of x such that P(X < x) = 0.05 and 0.95.

Use those values with xline to add the vertical lines to the plot.

[Andi]

I attempted but did not work in my case, may you please provide a bit more detail. Thank you!

[Paul]

Why not show the code?

Here is an example based on the doc

SixMPG = [13;15;23;29;32;34];
pdSix = fitdist(SixMPG,'Kernel','Width',4);
x = 0:.01:45;
ySix = pdf(pdSix,x);
plot(x,ySix,'k-','LineWidth',2)
hold on
xvals = icdf(pdSix,[0.05 .95])

xvals = 1×2

9.7202 37.4349

xline(xvals,'r')

% verify xvals (approximately)
trapz(x(x<xvals(1)),ySix(x<xvals(1)))

ans = 0.0499

trapz(x(x<xvals(2)),ySix(x<xvals(2)))

ans = 0.9498

[Andi]

Thanks a lot this works for my case, i jsut wounder how can i specify the kernel type e.g. normal or gaussian kernal along with the bandwidth option.

[Paul]

According to fitdist

pdSix = fitdist(SixMPG,'Kernel','Kernel,',kernel,'Width',4);

where kernel is one of:

• 'normal'
• 'box'
• 'triangle'
• 'epanechnikov'

[Andi]

Thankl you very much !