Percentiles without Statistics Toolbox
24 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
ava mazaheri
am 26 Nov. 2020
Kommentiert: MikaelTulldahlCPAC
am 5 Mär. 2021
Hi,
I have a bunch of histograms from which I need to extract some percentiles (10th, 50th and 90th).
I know about the prctile function, but it requires the Statistics Toolbox in Matlab which I do not have. The only toolbox I have is the Signal Processing Toolbox.
Is there a way of obtaining percentiles other than the prctile function?
//AM
0 Kommentare
Akzeptierte Antwort
Star Strider
am 26 Nov. 2020
Try this:
x = 3*randn(1,100)+5; % Create Data
figure
yyaxis left
h = histogram(x, 20);
hold on
barvals = h.Values;
edgs = h.BinEdges;
ctrs = edgs(1:end-1) + (diff(edgs)/2); % Calculate Bar Center Locations
plot(ctrs, barvals, '+r')
hold off
ylabel('Number')
auc = 100*cumsum(barvals)/sum(barvals)+(0:numel(barvals)-1)*eps*100; % Cumulative Area
prctls = [10, 50, 90]; % Desired Percentiles
prctlctrs = interp1(auc, ctrs, prctls); % Percentile X-Locations
yyaxis right
plot(prctlctrs, prctls, 'dg', 'MarkerFaceColor','g') % Plot Percentiles
hold on
plot(ctrs, auc, '--k') % Plot Area (Optional)
hold off
grid
ylim([0 100])
xlabel('X')
ylabel('Cumulative Percent')
text(prctlctrs, prctls, sprintfc('%2d^{th} Percentile = %5.2f\\rightarrow ', [prctls; prctlctrs].'), 'HorizontalAlignment','right', 'VerticalAlignment','middle')
legend('Histogram', 'Centres', 'Percentiles', 'Area', 'Location','E')
.
3 Kommentare
MikaelTulldahlCPAC
am 5 Mär. 2021
if anyone needed percentiles unrelated to histograms, it's easy to calculate:
samplePoints = [0.1, 0.5, 0.9];
data = randn(10,3) % 3 sets with 10 samples each
percentiles = interp1(linspace(0, 1, size(data,1)), sort(data), samplePoints);
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu Descriptive Statistics and Visualization finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!