根据百分比更改直方图的背景颜色答案

【问题标题】：Change background color of histogram according to a percentage根据百分比更改直方图的背景颜色
【发布时间】：2019-05-17 04:41:34
【问题描述】：

我有一个直方图，我想为分布在直方图中 16.5% 到 83.5% 的点之间的背景着色。

我该怎么做？如何找到这些点？

数据位于file - 一列值中。

h = histogram( file, 50 );

【问题讨论】：

请添加您想要的示例

标签： matlab histogram matlab-figure

【解决方案1】：

只是想添加另一个变体。使用直方图属性和 prctile 找到限制：

data = randn(100000,1);

% Start with the original histogram
figure;
h=histogram(data,50);
% Find the bin edges you received.
be=h.BinEdges;

% Find the limits where your percentile limits lie
y=prctile(data,[16.5 83.5]);

% However, percentile limits will not generally concide with your bin-limits, so this must be fudged.

% Option A: Adjust be, to lie on the percentiles.
% DYI

% Option B: Adjust your limits for a pretty plot

% Find which be indicies are closest to the desired limits.
vals=y(:); 
rv=be(:)';
diffs=bsxfun(@minus,vals, rv); % Finds differences to all be for all vals.
[~,inds]=min(abs(diffs),[],2); % Finds the minimum ones.
vals=rv(inds);                 % Find values to use for the cutoff.

% Replace the original plot with the inner cut.
h1=histogram(data(data>vals(1) & data<vals(2)),'BinEdges',be);
hold on;
% Plot the data outside the limits.
h2=histogram(data(data<vals(1) | data>vals(2)),'BinEdges',be);

% Pretty colors have ensued. As per post, you can color the tails to
% something else
h2.FaceColor='white';

感谢 Tom R 四舍五入到特定值： https://se.mathworks.com/matlabcentral/fileexchange/37674-roundtowardvec

【讨论】：

很好的答案
@AnderBiguri，谢谢，我很感激！
非常感谢，vals是什么意思？
vals 只是四舍五入到所需 bin 限制的“值”的缩写。关于ax=gca; ax.FontSize=12;这对我来说很好（R2016b 版本）。检查您是否将正确的轴加载到 ax 中。
不错的结果，值得注意的是prctile 依赖于统计和机器学习工具箱。

【解决方案2】：

详情请看代码cmets，基本上你可以用patch来高亮背景，还有一些逻辑索引来找出哪些bin在你的16.5% - 83.5%的阈值之内。

这使用bar 和histcounts 创建直方图而不是histogram，因为您可以获得更多有用的输出，并且我们需要在绘图之前执行中间步骤。

rng(0); % for repeatable random numbers
x = normrnd( 0, 1, 1000, 1 ) * 10; % Create data

% Get the histogram counts with 50 bins
[hc, edges] = histcounts( x, 50 );

% Lower and upper bounds we're interested in highlighting
region = [0.165, 0.835];
% Cumulative percentage across the bins
pct = cumsum( hc ) / sum( hc );
% Index to get which meet our bounds
idx = pct >= region(1) & pct <= region(2);

% Set up the plot
x = (edges(1:end-1)+edges(2:end))/2;
maxY = 1.1*max(hc);
n = nnz(idx);

% Plot
figure; hold on
patch( [x(idx),fliplr(x(idx))], [zeros(1,n),ones(1,n)]*maxY, 'y', 'edgecolor', 'none' );
bar( x, hc );
hold off
ylim( [0, maxY] );

结果：

【讨论】：