Binning 2 columns and summing 3rd

I have three very large datasets of latitude, potential temperature and ozone. I want to binning the data of latitude and potential temperature and according to the bins, I want to sum ozone data and put it into each bins that I have created. I have tried using accumaray which shows me,

"Error using accumarray

First input SUBS must contain positive integer subscripts."

Indeed ozone data contains float numbers.

Am I missing something or what I could not understand.

I am attaching script as under (also pasting the copied script):

I have also attached mat files of the data.

clc;

clear all;

close all;

ozonef = cell2mat(ozone);

poslatf = cell2mat(poslat);

Tpotf = cell2mat(Tpot);

% % %bining the data-

poslatbin = -90:2:90;

Tpotbin = 250:5:380; %%%% I have values in between this range

nposlat = length(poslatbin);

nTpot = length(Tpotbin);

PL = discretize(poslatf, poslatbin);

TP = discretize(Tpotf, Tpotbin);

ozonef(isnan(ozonef))=min(ozonef);

idx = sub2ind([nposlat nTpot], PL, TP);

FO = accumarray(idx, ozonef, [nposlat*nTpot 1], @(x) mean(x));

contourf(PL, TP, FO);

Answer 1

[Cris LaPierre]

I would put your data into a table, then use groupsummary to bin and sum the data as you describe. You will find the Access Data in Tables page helpful.

load latitude
load temperature
load ozone

% create table
data = table(latitude,temperature,ozone)

data = 47101×3 table
    latitude    temperature    ozone
    ________    ___________    _____

     24.876       290.97       1.41 
     24.866       290.92       1.41 
     24.856       291.11       1.41 
     24.845       291.27       1.41 
     24.833       291.28       1.41 
     24.825       291.13       1.41 
     24.815       291.24       1.41 
     24.805       291.43       1.41 
     24.795        291.5       1.41 
     24.785       291.56       1.41 
     24.775       291.52       1.41 
     24.765       291.43       1.41 
     24.755       291.54       1.41 
     24.746       291.75       1.41 
     24.736        291.8       1.41 
     24.726       292.03       1.41 

% define bins
poslatbin = -90:2:90;      
Tpotbin = 250:5:380;

% create table of binned data
binTbl = groupsummary(data,["latitude","temperature"],{poslatbin,Tpotbin},'sum',"ozone")

binTbl = 370×4 table
    disc_latitude    disc_temperature    GroupCount    sum_ozone 
    _____________    ________________    __________    __________

     [-34, -32)        [330, 335)             36            15973
     [-34, -32)        [335, 340)             56            21635
     [-34, -32)        [340, 345)            743            52250
     [-34, -32)        [345, 350)            455            45569
     [-34, -32)        <undefined>           497            30712
     [-32, -30)        [335, 340)             44            16214
     [-32, -30)        [340, 345)            517       1.3762e+05
     [-32, -30)        [345, 350)            648            78325
     [-32, -30)        <undefined>           351            19507
     [-30, -28)        [335, 340)             39            11469
     [-30, -28)        [340, 345)            953       1.1891e+05
     [-30, -28)        [345, 350)            475            50072
     [-30, -28)        <undefined>           195            18177
     [-28, -26)        [335, 340)            332            16639
     [-28, -26)        [340, 345)           1003            62685
     [-28, -26)        [345, 350)            540            51063

[Cris LaPierre]

I looked at your m-file. It looks like the desired end result is a filled contour plot. To achieve that, you have to make some adjustments to the settings in groupsummary. By default, it does not return the results for empty and missing bins (for example, you have no data in the [-90,-88) bin. To create a contour plot, you need data for every bin. Here's what that code might look like.

% Start the same as before
load latitude
load temperature
load ozone
data = table(latitude,temperature,ozone);
poslatbin = -90:2:90;          
Tpotbin = 250:5:380;  

% The two new settings will include data for every bin
binTbl = groupsummary(data,["latitude","temperature"],{poslatbin,Tpotbin},'sum',"ozone",...
    "IncludeEmptyGroups",true,"IncludeMissingGroups",true)

binTbl = 2457×4 table
    disc_latitude    disc_temperature    GroupCount    sum_ozone
    _____________    ________________    __________    _________

     [-90, -88)         [250, 255)           0             0    
     [-90, -88)         [255, 260)           0             0    
     [-90, -88)         [260, 265)           0             0    
     [-90, -88)         [265, 270)           0             0    
     [-90, -88)         [270, 275)           0             0    
     [-90, -88)         [275, 280)           0             0    
     [-90, -88)         [280, 285)           0             0    
     [-90, -88)         [285, 290)           0             0    
     [-90, -88)         [290, 295)           0             0    
     [-90, -88)         [295, 300)           0             0    
     [-90, -88)         [300, 305)           0             0    
     [-90, -88)         [305, 310)           0             0    
     [-90, -88)         [310, 315)           0             0    
     [-90, -88)         [315, 320)           0             0    
     [-90, -88)         [320, 325)           0             0    
     [-90, -88)         [325, 330)           0             0    

% need to rearrange the data into a nxm matrix
% rows correspond to y (poslatbin), and columns to x (Tpotbin)
cdata = reshape(binTbl.sum_ozone,length(poslatbin),length(Tpotbin));
cdata(cdata==0)=min(ozone); % Replace 0s with min ozone value (might not be necessary)

% Create your contour plot
contourf(Tpotbin,poslatbin,cdata)
colorbar

[Vaidehi Joshi]

Thank you so much. I appreciate your help.

It worked, but I want to plot this data. I do not know how to plot data, it seems complex to me. I want contour plot as I mentioned above in my script.

[Vaidehi Joshi]

Thanks a lot for your help. It is working.

[Vaidehi Joshi]

I have a one question here. I am changing X and Y axis (in my case right now it is temperature and latitude,respectivley), matrix changes too but values would be false. how can I check what we have done here are the correct values or not?

[Cris LaPierre]

I don't understand the question. My suggestion - make the changes, and just be sure to update the corersponding variable names in the code I shared, and see if it works.

[Vaidehi Joshi]

I did change the axises. As I made the changes in these lines

cdata = reshape(binTbl.sum_ozone,length(Tpotbin instead of poslatbin),length(postlatbin instead of Tpotbin)); and

contourf(Tpotbin,poslatbin,cdata) same as above.

cdata matrix will be changed right!

Here what I noticed with the reshape function is, values are also changing, accordingly plot.

I want to crosscheck whether these values are correct or not.