How to determine average and standard deviation of y axis values for corresponding bin of x axis?

Sir i have two data series say

a                  b
25.36737061 -27.47956892
20.54391479 -23.68162398
16.76391602 -16.65254461
9.47177124 -19.20600915
16.25158691 -18.56570783
4.462646484 -14.39363913
7.785919189 -14.98048449
12.27481079 -18.49125231
4.851806641 -19.91135093
2.111236572 -5.049334665
-1.457702637 -6.51219601
1.85055542 -1.299793246

and i want to plot for bin width of 'a' (say 5,10,15 etc) with the corresponding average and standard deviation of b. Please help me with a easy way since the series are very large?

[trailokya]

if i do

[~,idx] = histc(a, 0:5:50);
result = accumarray(idx, b, [], @mean);

The following error is coming Error using accumarray First input SUBS must contain positive integer subscripts.

What should i do sir, any help?

[Kelly Kearney]

histc will return 0 for the idx value of any points that don't fall within your bins (i.e. <0 or >50). You could filter those points out, or change the idx value for those points:

[~,idx] = histc(a, 0:5:50);
idx(idx == 0) = max(idx)+1;
bavg = accumarray(idx, b, [], @mean);

Note that bavg(1:end-2) now corresponds to the averages within the bins, bavg(end-1) is the average of values exactly equal to your rightmost bin edge (50), and bavg(end) is the average of all values either <0 or >50. I usually prefer to make sure that my bins fully encompass the data values... easier to analyze that way.

[Image Analyst]

The average of the bins is not the average of the original data. Be sure you know what you really want and compute it correctly.

[Star Strider]

Context: http://www.mathworks.com/matlabcentral/answers/243174-how-to-bin-for-scatter-plot-from-two-data-series.

I explained this in my Answer to your previous post (that you have as yet to Accept).

[Kelly Kearney]

To simplify the tracking of indices, you may want to look at this aggregatehist.m function... it's basically a wrapper around the histc plus accumarray method demonstrated by Star Strider and myself.

ab = [...
25.36737061 -27.47956892
20.54391479 -23.68162398
16.76391602 -16.65254461
9.47177124 -19.20600915
16.25158691 -18.56570783
4.462646484 -14.39363913
7.785919189 -14.98048449
12.27481079 -18.49125231
4.851806641 -19.91135093
2.111236572 -5.049334665
-1.457702637 -6.51219601
1.85055542 -1.299793246];

bin = -5:5:30;
[aa,bb] = aggregatehist(bin, ab(:,1), ab(:,2));
dev = cellfun(@std, bb);
avg = cellfun(@mean, bb);
amid = (bin(1:end-1)+bin(2:end))./2;

plot(ab(:,1), ab(:,2), '.');
hold on;
errorbar(amid, avg, dev, 'linestyle', 'none', 'marker', 'o');
set(gca, 'xtick', bin, 'xgrid', 'on');

Answer 1

[Image Analyst]

Why do you say (the badly-named) "a" is the bin width and then say it has values of 5, 10, or 15 when it clearly does not have any of those values? You can specify the edges of the bins so that the bin widths are exactly what you want. Like if you want a bin width of a(1), which is 25.3673, then you can do

edges = 0 : a(1) : 50

and then pass edges into histc() or histcounts().

[trailokya]

Actually i need the mean and standard deviation of y for a particular range of x say <0, 0-5,5-10, 10-15 etc.

[trailokya]

here a is x values and b is y values

[Image Analyst]

% Find indexes where x is in the range 0-5.
indexesInRanges = x>0 & x<5;
% Get mean of y for those indexes:
meanY = mean(y(indexesInRanges));
% Get standard deviation of y for those indexes:
stdY = std(y(indexesInRanges));

[Wade Mao]

You're the GOD of matlab! I love you Image Analyst !!

[Image Analyst]

Thanks Wade. Maybe so, since I seem to be immortal, what with my responses of long ago living forever. You can "vote" for answers if you want to give anybody additional "reputation points".