Trying to remove Nans when plotting histogram, pdf and cdf
显示 更早的评论
clear;
load InsulinReadings.mat
xX2 = InsulinReadings;
xX2(xX2==0)=missing;
A2 = mean(xX2,'all',"omitnan")
B2 = median(xX2,'all',"omitnan")
C2 = max(xX2,[],'all',"omitnan")
D2 = min(xX2,[],'all', "omitnan")
figure
histogram(InsulinReadings(~isnan(InsulinReadings),128,'Normalization')
xlabel('Insulin ng/dL')
%Now get pdf
[D PD] = allfitdist(xX2,'PDF');
xlabel('Insulin ng/dL');
%Now get the CDF
[D PD] = allfitdist(xGlucoseReadings,'CDF');
xlabel('Insulin ng/dL')
采纳的回答
I think you basically have it right. I just "fixed" a syntax error on the line where you call histogram(). ("fixed" is in quotes because I can't be sure what you're going for there.)
(Also, looks like allfitdist.m has been removed from the File Exchange, so I can't run it, but maybe your copy does the right thing here - I don't know.)
clear;
load InsulinReadings.mat
xX2 = InsulinReadings;
xX2(xX2==0)=missing;
A2 = mean(xX2,'all',"omitnan")
B2 = median(xX2,'all',"omitnan")
C2 = max(xX2,[],'all',"omitnan")
D2 = min(xX2,[],'all', "omitnan")
figure
histogram(InsulinReadings(~isnan(InsulinReadings)),128)%,'Normalization')
xlabel('Insulin ng/dL')
%Now get pdf
[D PD] = allfitdist(xX2,'PDF');
xlabel('Insulin ng/dL');
%Now get the CDF
[D PD] = allfitdist(xGlucoseReadings,'CDF');
xlabel('Insulin ng/dL')
6 个评论
Nathaniel Porter
2022-2-25
I see.
Well, they're not necessarily zeros, they're just values that fall within the first bin of the histogram (i.e., closer to zero than to the right edge of the second bin). I guess it makes sense to plot a histogram of xX2 rather than InsulinReadings since xX2 has the zeros replaced with NaNs (but you still get a lot of small values).
If you want to remove any value that would be in the first bin of the histogram, you have to know what your histogram edges are before creating the histogram (which is ok), or you could replace any value less than some threshold with a NaN (or a 0, then replace all the 0s with NaNs like you're already doing), but then you have to choose some threshold, so your histogram will depend on that choice.
clear;
load InsulinReadings.mat
xX2 = InsulinReadings;
nnz(xX2 == 0) % 41209 zeros
nnz(isnan(xX2)) % 0 NaNs
xX2(xX2==0)=missing;
nnz(isnan(xX2)) % now 41209 NaNs
figure
h = histogram(xX2(~isnan(xX2)),128);%,'Normalization')
xlabel('Insulin ng/dL')
% now make a new histogram with values in the first bin replaced with NaNs
edges = get(h,'BinEdges');
xX2(xX2 < edges(2)) = NaN;
figure();
h = histogram(xX2(~isnan(xX2)),128);%,'Normalization')
xlabel('Insulin ng/dL')
Nathaniel Porter
2022-2-25
Nathaniel Porter
2022-2-25
There probably are built-in functions, but I don't know what they are off the top of my head (maybe search the documentation).
It's relatively straighforward to calculate a PDF and CDF from the properties of the histogram:
clear;
load InsulinReadings.mat
xX2 = InsulinReadings;
xX2(xX2==0)=missing;
figure();
h = histogram(xX2(~isnan(xX2)),128);%,'Normalization')
% now make a new histogram with values in the first bin replaced with NaNs
edges = get(h,'BinEdges');
xX2(xX2 < edges(2)) = NaN;
figure();
h = histogram(xX2(~isnan(xX2)),128);%,'Normalization')
xlabel('Insulin ng/dL')
% pdf and cdf
figure();
edges = get(h,'BinEdges');
counts = get(h,'BinCounts');
bin_centers = (edges(1:end-1)+edges(2:end))/2;
total_counts = sum(counts);
pdf = counts/total_counts;
cdf = cumsum(counts)/total_counts;
plot(bin_centers,pdf,'LineWidth',2);
hold on
plot(bin_centers,cdf,'LineWidth',2);
xlim(bin_centers([1 end]));
legend('PDF','CDF');
Nathaniel Porter
2022-2-26
更多回答（0 个）
类别
在 帮助中心 和 File Exchange 中查找有关 Data Distribution Plots 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
