boxplot bins
in [Matlab]
|
From: Dustin on 12 Aug 2010 09:44 I think this is a fairly straight forward question but well see. I have an arbitrarily sized data set x and y, in two column vectors. I want to do box plots of y for even intervals of x on the same figure. How can I do this without breaking up my data set? thanks.
From: Kelly Kearney on 12 Aug 2010 10:40 "Dustin " <dbrisset(a)gmail.com> wrote in message <i40tr7$3j9$1(a)fred.mathworks.com>... > I think this is a fairly straight forward question but well see. > > I have an arbitrarily sized data set x and y, in two column vectors. I want to do box plots of y for even intervals of x on the same figure. > > How can I do this without breaking up my data set? > > thanks. Use histc to group your data by interval, and use boxplot with the group input: x = rand(100,1)*100; y = rand(size(x)); edge = 0:10:100; [n,bin] = histc(x,edge); boxplot(y, bin) You may want to play around with the positions, widths, and xlabels of each boxplot so it better reflects the actual intervals being displayed. -Kelly
From: Dustin on 12 Aug 2010 11:34 "Kelly Kearney" <kakearney(a)nospamgmail.com> wrote in message <i4114s$c8s$1(a)fred.mathworks.com>... > "Dustin " <dbrisset(a)gmail.com> wrote in message <i40tr7$3j9$1(a)fred.mathworks.com>... > > I think this is a fairly straight forward question but well see. > > > > I have an arbitrarily sized data set x and y, in two column vectors. I want to do box plots of y for even intervals of x on the same figure. > > > > How can I do this without breaking up my data set? > > > > thanks. > > Use histc to group your data by interval, and use boxplot with the group input: > > x = rand(100,1)*100; > y = rand(size(x)); > edge = 0:10:100; > [n,bin] = histc(x,edge); > boxplot(y, bin) > > You may want to play around with the positions, widths, and xlabels of each boxplot so it better reflects the actual intervals being displayed. > > -Kelly This is super close to what I want, but is great help for me. Let me add a little more detail to what I was going for. I want to transpose this on top of the scatter plot of (x,Y) corresponding to where the data lies. I don't think this is quite grabbing the bins appropriately. I'll try to elaborate a little better. Looking at my code below... This appears to be breaking my x-range of data into 50 bins. where i want a specified x-range (0-25) broken into 50 bins. Theres something I'm not quite understanding, and the help isn't very helpful to me. It also keeps hijacking my x-tick marks but i can probably handle this. scatter(data.x,data.y) hold on edge = 0:.5:25; [~,bin] = histc(data.x,edge); boxplot(data.y, bin)
From: Tom Lane on 12 Aug 2010 13:40 > I'll try to elaborate a little better. Looking at my code below... This > appears to be breaking my x-range of data into 50 bins. where i want a > specified x-range (0-25) broken into 50 bins. Theres something I'm not > quite understanding, and the help isn't very helpful to me. > It also keeps hijacking my x-tick marks but i can probably handle this. > > scatter(data.x,data.y) > hold on > edge = 0:.5:25; > [~,bin] = histc(data.x,edge); > boxplot(data.y, bin) Type "help histc" and you'll see that the bin number is 0 for out-of-range values. The boxplot function doesn't treat the value 0 as anything special. If you want to exclude bin 0, you might try xmeans = grpstats(data.x,bin) boxplot(data.y(bin>0), bin(bin>0),'position',xmeans(2:end)) Here I've also positioned the boxes at the means of the x values in each bin. You may need to play with this some. For example, I know I have values in bin 0 so I omit the first xmeans value. You might need to do that only conditionally. -- Tom
From: Kelly Kearney on 12 Aug 2010 13:40
"Dustin " <dbrisset(a)gmail.com> wrote in message <i4149j$aao$1(a)fred.mathworks.com>... > "Kelly Kearney" <kakearney(a)nospamgmail.com> wrote in message <i4114s$c8s$1(a)fred.mathworks.com>... > > "Dustin " <dbrisset(a)gmail.com> wrote in message <i40tr7$3j9$1(a)fred.mathworks.com>... > > > I think this is a fairly straight forward question but well see. > > > > > > I have an arbitrarily sized data set x and y, in two column vectors. I want to do box plots of y for even intervals of x on the same figure. > > > > > > How can I do this without breaking up my data set? > > > > > > thanks. > > > > Use histc to group your data by interval, and use boxplot with the group input: > > > > x = rand(100,1)*100; > > y = rand(size(x)); > > edge = 0:10:100; > > [n,bin] = histc(x,edge); > > boxplot(y, bin) > > > > You may want to play around with the positions, widths, and xlabels of each boxplot so it better reflects the actual intervals being displayed. > > > > -Kelly > > This is super close to what I want, but is great help for me. Let me add a little more detail to what I was going for. I want to transpose this on top of the scatter plot of (x,Y) corresponding to where the data lies. I don't think this is quite grabbing the bins appropriately. > > I'll try to elaborate a little better. Looking at my code below... This appears to be breaking my x-range of data into 50 bins. where i want a specified x-range (0-25) broken into 50 bins. Theres something I'm not quite understanding, and the help isn't very helpful to me. > > It also keeps hijacking my x-tick marks but i can probably handle this. > > scatter(data.x,data.y) > hold on > edge = 0:.5:25; > [~,bin] = histc(data.x,edge); > boxplot(data.y, bin) I think it's doing what you want it to do, but you need to give it x positions and widths to position the boxes in the correct area. The below should do that, as well as correct the annoying hijacking of the x-ticks (I hate that boxplot does that). Also, unless you plan to use varying colors and sizes in your scattered data, plot is a lot faster than scatter. x = rand(1000,1) .* 25; y = sin(x/4) + randn(size(x)); edge = 0:.5:25; [n,bin] = histc(x, edge); xmid = (edge(1:end-1) + edge(2:end))./2; dx = diff(edge); figure; plot(x,y,'k.'); hold on; boxplot(y, bin, 'positions', xmid, 'widths', dx); set(gca, 'xticklabelmode', 'auto', 'xtick', 0:5:25); |