Mean error for binned values

Hello guys!

I have a data sets with the measurements and GPS coordinates, where the values were measured. For this data set I am using a bins of 10 meters and 20 meters. I am calculating a means from the values.

[latbin, lonbin] = hista(w_files{ii}.lat,w_files{ii}.lon,0.0001);% bin at 10x10m,hista binning + computing bins
[w_files{ii}.latEq, w_files{ii}.lonEq] = grn2eqa(w_files{ii}.lat,w_files{ii}.lon);% Convert coordinates to equidistant cartesian coordinates
[latbinEq, lonbinEq] = grn2eqa(latbin, lonbin);% Convert coordinates to equidistant cartesian coordinates
dist = pdist2([w_files{ii}.lonEq,w_files{ii}.latEq],[lonbinEq, latbinEq]);% Compute distance between each coordinate and each bin-center
[~, w_files{ii}.bin10] = min(dist,[],2);% Add bin ID numbers to table

%% making a medians from  bin variable values 
time_b10 = splitapply(@median,w_files{ii}.time, w_files{ii}.bin10);
lon_b10 = lonbin;
lat_b10 = latbin;
lon_b10 = splitapply(@nanmedian,w_files{ii}.lon, w_files{ii}.bin10);
lat_b10 = splitapply(@nanmedian,w_files{ii}.lat, w_files{ii}.bin10);
sinr_lte_b10 = splitapply(@nanmedian,w_files{ii}.sinr_lte, w_files{ii}.bin10);
cinr_lte_b10 = splitapply(@nanmedian,w_files{ii}.cinr_lte, w_files{ii}.bin10);
rsrp_lte_b10 = splitapply(@nanmedian,w_files{ii}.rsrp_lte, w_files{ii}.bin10);
prb_dl_lte_b10 = splitapply(@nanmedian,w_files{ii}.prb_dl_lte, w_files{ii}.bin10);
prb_ul_lte_b10 = splitapply(@nanmedian,w_files{ii}.prb_ul_lte, w_files{ii}.bin10);
bler_dl_lte_b10 = splitapply(@nanmedian,w_files{ii}.bler_dl_lte, w_files{ii}.bin10);
bler_ul_lte_b10 = splitapply(@nanmedian,w_files{ii}.bler_ul_lte, w_files{ii}.bin10);
tp_mac_d_lte_b10 = splitapply(@nanmedian,w_files{ii}.tp_mac_d_lte, w_files{ii}.bin10);
tp_mac_u_lte_b10 = splitapply(@nanmedian,w_files{ii}.tp_mac_u_lte, w_files{ii}.bin10);
tp_pdsch_lte_b10 = splitapply(@nanmedian,w_files{ii}.tp_pdsch_lte, w_files{ii}.bin10);
tp_pusch_lte_b10 = splitapply(@nanmedian,w_files{ii}.tp_pusch_lte, w_files{ii}.bin10);

This was working perfectly for me for last few weeks but now I got a data set where I am unable to use this 10x10 bin. It gives me those two errors:

Error using splitapply (line 111)

For N groups, every integer between 1 and N must occur at least once in the vector of group numbers.

Error in b10 (line 37)

time_b10 = splitapply(@median,w_files{ii}.time, w_files{ii}.bin10);

The odd think is that in this particular case I have two datasets which has same number of rows and same variables just the measured values are different. One works as intended and second one not.

Same If I try to run 20x20 bin its works flawlessly but when I run the 10x10m variation it gets stuck.

Only difference in the codes is naming ( xxx_B10 / xxx_b20) and this one line of code:

[latbin, lonbin] = hista(w_files{ii}.lat,w_files{ii}.lon,0.0001);% bin at 10x10m,hista binning + computing bins

[latbin, lonbin] = hista(w_files{ii}.lat,w_files{ii}.lon,0.0004);% bin at 20x20m,hista binning + computing bins

If you wanna know a bit more of my code please folow: https://www.mathworks.com/matlabcentral/answers/832518-geographical-binning-and-evaluation-of-results?s_tid=mlc_ans_email_view#comment_1530173

Will appreciate any idea why I am experiencing this difficulties or how to solve them. Please try to explain the ideas as thorough as possible, because I think that I might be dumb.

Answer 1

[Steven Lord]

Rather than calling splitapply multiple times, I'd call groupsummary once and have it operate on multiple data variables at once. Or you might need two groupsummary calls, once with @median to operate on the times and once with @(x) median(x, 'omitnan') to operate on the other variables, if the missing data in the times is important. That way you only have to specify the groupbins once.

[Karel Starý]

Can you please give me a specific example how to proceed in my case because I read all info around the groupsummary and I did not understood it.