CODEWITHSUNDEEP
matlab
HCAI
HCAI

Reputation: 2263

Errorbars on bar-graph

I'm trying to get: a nice bar graph of groups. I'd like error bars on each bar like this (or similar at your discretion):

enter image description here

I have the means: 

ff =
             Medical eq    Patient   Hygiene   Near bed   Far bed
Direct        1.2759    0.9253    0.6092    1.0460    1.3276
Housekeeping  1.0833    0.2500    0.5833    2.1667    1.3333
Mealtimes      0        0.3000    0.7000    1.4500    0.5000
Medication    0.3694    0.7838    0.8919    1.5495    1.0541
Misc          0.1059    0.1294    0.4118    0.8000    0.8353
Personal      0.0370    0.4074    0.8148    1.2593    0.7037

the standard deviations

ee =

2.0411    1.1226    0.8378    1.5007    1.3776
1.3114    0.4523    0.6686    2.4058    1.1547
     0    0.7327    1.3803    2.1392    0.6070
0.7499    0.9186    1.0300    1.2844    1.3062
0.4371    0.3712    0.7605    1.0212    0.6699
0.1925    0.6939    1.6417    3.5582    1.5644

instead I get this:

enter image description here

by using:

bb=bar(ff'); hold all
data=repmat([1:6]'*ones(5,1)',1,1)
er=errorbar(data, ff, ee, '.')

I read I have to find the centre of each bar? That's crazy! Any workaround?

Upvotes: 3

Views: 17917

Answers (2)

Gerbrand Visscher
Gerbrand Visscher

Reputation: 351

the centre of each bar can be obtained by 

x = get(get(h(i),'children'),'xdata');
barsx=mean(x,1);

barsx gives the center for every ith element of every bar subset.

h=bar(bars)
for i=1:6
  x = get(get(h(i),'children'),'xdata');
  barsx(1:6,i)=mean(x,1)
end
hold all
h=errorbar(barsx,bars,barsvar)

to have the errors having the same colors as bar:

figure()
h=bar(bars) 
col=[0 0 1;0 1 0;1 1 0; 1 1 1; 0 0 0; 0 1 1];
colormap([0 0 1;0 1 0;1 1 0; 1 1 1; 0 0 0; 0 1 1])
hold all
for i=1:6
  x = get(get(h(i),'children'),'xdata')
  barsx=mean(x,1)
  h1=errorbar(barsx',bars(1:6,i),barsvar(1:6,i),'color',col(i,:))
  set(h1,'linestyle','none')
end

Upvotes: 3

Junuxx
Junuxx

Reputation: 14261

It's really not that crazy!

You can just use errorbar with the data that you have readily available.

This code computes the correct locations, by adding 1 for every group and 1/7 for each bar within a group.

for i = 1:5
    j = 1:6; 
    x = -0.5 + i + 1/7 * j; 
    errorbar(x, ff(j,i), ee(j,i), '.'); 
end

Results in:

enter image description here

(Leaving out the labels, but otherwise this seems pretty similar to what you were looking for)

Upvotes: 2

Related Questions