Extracting multiple row values (with unequal range) in a column separately

I have a csv file 28526 x 1. These are a single variable for 45 years. I want to find the mean of 45 years (45 x1) seperately. But all 45 years unequal ranges. How to find the mean?

-----------------------------------------------------------------------------------

clc;clear all

a = readmatrix('His_wind.csv');

b = a(20279:20981, 1);

b(b == 9999) = NaN;

c = b/10;

d = mean(c,'omitnan');

--------------------------------------------------------------------------------------

Here I am selecting each range of values and calculating mean seperately for each year. Please advice on how to do it together.

[Jonas]

how do you know the true range for each year? i doubt, can you provide the data file?

[ANIE]

please find the data file attached. The header P DEW seperates each year

Answer 1

[Mathieu NOE]

hello

try this :

clc;clear all
a = readmatrix('saf_wind.csv');
ind_header = find(isnan(a)); % lines with 'P DEW' (seperates each year)

% this for loop will do the yearly averaging on data between first and last
% detected lines with 'P DEW' - data before the first P DEW and data after
% the last P DEW mark are not computed
for ci =1:numel(ind_header)-1
    start = ind_header(ci);
    stop = ind_header(ci+1);
    data = a(start:stop);
    data(data > 9000) = NaN; % remove the 9999 from data (using > 9000 instead of == 9999 for potential rounding issues
    d(ci) = mean(data/10,'omitnan');
end

plot(d)

[Mathieu NOE]

FYI

an equivalent code without NaN acrobatics

clc;clear all
a = readmatrix('saf_wind.csv');
ind_header = find(isnan(a)); % lines with 'P DEW' (seperates each year)

% this for loop will do the yearly averaging on data between first and last
% detected lines with 'P DEW' - data before the first P DEW and data after
% the last P DEW mark are not computed

for ci =1:numel(ind_header)-1
    start = ind_header(ci)+1;
    stop = ind_header(ci+1)-1;
    data = a(start:stop);
    data(data > 9000) = []; % remove the 9999 from data (using > 9000 instead of == 9999 for potential rounding issues
    d(ci) = mean(data/10);
end

plot(d)

[Mathieu NOE]

Just figured out as you wanted the results for the 45 years, this means I had to add the computations for the data before the first and after the last mark 'P DEW'

this is it :

clc;clear all
a = readmatrix('saf_wind.csv');
ind_header = find(isnan(a)); % lines with 'P DEW' (seperates each year)

% this for loop will do the yearly averaging on data between first and last
% detected lines with 'P DEW' - data before the first P DEW and data after
% the last P DEW mark are not computed

for ci =1:numel(ind_header)-1
    start = ind_header(ci)+1;
    stop = ind_header(ci+1)-1;
    data = a(start:stop);
    data(data > 9000) = []; % remove the 9999 from data (using > 9000 instead of == 9999 for potential rounding issues
    dd(ci) = mean(data/10);
end

% add the computation for the data before the first detected line with 'P DEW'
    start = 1;
    stop = ind_header(1)-1;
    data = a(start:stop);
    data(data > 9000) = []; % remove the 9999 from data (using > 9000 instead of == 9999 for potential rounding issues
    d_first = mean(data/10);

% add the computation for the data afetr the last detected line with 'P DEW'
    start = ind_header(end)+1;
    data = a(start:end);
    data(data > 9000) = []; % remove the 9999 from data (using > 9000 instead of == 9999 for potential rounding issues
    d_last = mean(data/10);    
    
% concatenation
d_final = [d_first dd d_last];

[ANIE]

Thanks a lot. It worked.

[Mathieu NOE]

My pleasure !

Answer 2

[Jonas]

dat=readmatrix('saf_wind.csv');
dat(dat==9999)=[];
whereIsNan=find(isnan(dat));
yourMeans=zeros(1,numel(whereIsNan)+1);
for partNr=1:numel(whereIsNan)+1
    
    if partNr==1
        currVec=dat(1:whereIsNan(partNr(1))-1);
    elseif partNr==numel(whereIsNan)+1
        currVec=dat(whereIsNan(end)+1:end);
    else
        currVec=dat(whereIsNan(partNr-1)+1:whereIsNan(partNr)-1);
    end
    yourMeans(partNr)=mean(currVec);
end

[ANIE]

Thanks a lot. This also works. Only I had to divide it by 10 as i needed the matrix in fraction.