Question

Lets say we have following vector:

data=a=[2.3 3.2 4.1 6.2 7.3 6.4 5.5 4.3 3.2 2.6 1.7 3.4 4.5 5.7 6.8];

If we count the numbers only by using the hist-function, we get something like that:

[n xout]=hist(a,[1:1:max(a)])

n =

 0     0     2     4     3     1     3     2

xout =

 0     1     2     3     4     5     6     7

What i now want to make is to sum each element of a bin:

xout =

 0     1     2     3     4     5     6     7

n =

 0     0     2     4     3     1     3     2

binsum =

0      0     4    12.4 etc.

for the third bin for example i have n(3)=2 values that are between 1,5 and 2,5 (size of a bin):1.7 and 2.3 -> 1.7+2.3=4 -> binsum(3)=4
for the fourth bin i have n(4)=4 values that are between 2,5 and 3,5:3.2+3.2+3.4+2.6=12.4-> binsum(4)=12.4 etc.

is there a simple function which do this job?

Answer 1

If your histogram bins are defined by the array minBin:binWidth:maxBin, then you can find the indices of the bins, and then sum up the data like this:

minBin = 1;
binWidth = 1;
maxBin = 7;

data=[2.3 3.2 4.1 6.2 7.3 6.4 5.5 4.3 3.2 2.6 1.7 3.4 4.5 5.7 6.8];

%# to create index: First, take care of out-of-range data points
tmp = max(min( data, maxBin),minBin);
%# then, subtract minimum, divide by step, round: this is the binIdx
%# add 1 so that we don't start counting at 0
idx = round( (tmp-minBin)/binWidth ) +1;

%# now we can use accumarray to sum up the data
binSum = accumarray(idx(:),data(:),[floor((maxBin-minBin)/binWidth)+1, 1], @sum, 0)

binSum =

         0
    4.0000
   12.4000
    8.4000
    4.5000
   23.8000
   14.1000

Note that I don’t get two zeros first, because I use the bins you used as input to hist, not the ones you apparently used to generate the outputs. Also, you can get the counts by replacing data by ones(length(data),1).