Cell Function

I've created a sliding window using the cell function. I used this as I am creating the sliding window for a daily time series of data over a few hundred years. The sliding window is 30 years long and because of leap years each row is a different length. So I now have an array of data that looks like this:

data =

10957x1 double

10957x1 double

10958x1 double

and so on...

I'm unfamiliar with the cell function however as this code was shown to me. How do I now use the data in this series? Whenever I type the command for example:

data (1,1)= 10957x1 double

But I need to do a number of things such as take the mean of each vector in the array, how do I do this?

Answer 1

[Wayne King]

x = {randn(1000,1), randn(1000,1), randn(1000,1)};
meanz = cellfun(@mean,x,'UniformOutput',false);
meanz = cell2mat(meanz);

[Wayne King]

you did say that these vectors are stored in a cell array, right?

[James]

I need to do a few more things than just take the mean. That was just an example.

I need to get the data from each cell into a usable format so I can say for example;

for 1:n

mean(data(n,1))

end

and

for 1:n

detrend(data(n,1))

end

[James]

Can you explain how to do this?

[Wayne King]

you can use cellfun to do that.

[Jan]

@James: Use curly braces to access cell elements:

for 1:n, mean(data{n,1}), end

for 1:n, detrend(data{n,1}), end

Answer 2

[Andrei Bobrov]

data2 = cellfun(@(x){mean(x) detrend(x)},x,'un',0)

ADD

data2 = cellfun(@(x)[mean(x);detrend(x)],x,'un',0)
datadouble = [data2{:}];
datamean = datadouble(1,:)
datadetrend = datadouble(2:end,:);

[James]

This just gives me the same problem?

It still outputs it as a cell array, I cant do anything with a vector where all it shows me is:

1x2 double, for every row

How can I extract the data from this as it is not currently usable?

I need to plot graphs of this data using many different functions, not just the ones I've put on here.

Answer 3

[Wayne King]

You can use two calls to cellfun and then use cell2mat

    data1 = cellfun(@mean,x,'UniformOutput',false);
    data1 = cell2mat(data1);
    data2 = cellfun(@detrend,x,'UniformOutput',false);
    data2 = cell2mat(data2);

[Wayne King]

Notice that data2 contains the detrended data as column vectors, which you can plot()

plot(data2)

data1 is a row vector of means

[James]

Detrend outputs a new value for each value I have. So at the moment using cell2mat the function just seems to put all 30 year period into one single column. Is there a way to define where the data goes? I have tried this but it doesn't seem to work?

detrendcell = cellfun(@(T32)detrend(T32,'linear'),T32,'un',0);

% detrends the data in each cell

n = numel(detrendcell);

for i=1:n

detrendT(i) = cell2mat(detrendcell(i));

end

Answer 4

[Fangjun Jiang]

Since data is a cell array, you need to use {} to reference its element, like data{1}, data{2}, data{3}, etc.

data{1} will be a 10957x1 double array, to reference its element, use data{1}(1), data{1}(100), data{1}(10957), etc.

You need to understand these basics. See http://www.mathworks.com/help/techdoc/learn_matlab/f4-2137.html#f4-2245

Answer 5

[James]

%%Data Loading

close all

clear all

cet = load ('cet_1772_2009.asc', '-ascii'); % Loads CET from asc file.
year = cet(:,1);

temp = cet(:,4); 

day = cet(:,3);  

month = cet(:,2);

% Loads Data from CET asc file and sets vectors

%%Constants and vectors

dates = datenum([year,month,day]);

%%30 Year Periods

StartYear=1772;

EndYear=2009;

Period=30;

T32=cell(EndYear-StartYear-Period+1,1);

for Year=StartYear:(EndYear-Period)

      StartCount=datenum(Year,1,1)-datenum(StartYear,1,1)+1;

      DataCount=datenum(Year+Period,12,31)-datenum(Year,12,31);

      T32{Year-StartYear+1}=temp(StartCount:(StartCount+DataCount),1);
end
%%Detrend, mean etc

meancell = cellfun(@(T32)mean(T32),T32,'un',0);

meanT = cell2mat(meancell);

detrendcell = cellfun(@(T32)detrend(T32,'linear'),T32,'un',0);

n = numel(detrendcell);

for i=1:n

n1 = numel(detrendcell{i});

detrendT(:,i) = cell2mat(detrendcell{i}(1:n1));

end

[James]

This is my code and it outputs the error:

??? Cell contents reference from a non-cell array object.

Error in ==> cell2mat at 43

cellclass = class(c{1});

Error in ==> Untitled2 at 63

detrendT(:,i) = cell2mat(detrendcell{i}(1:n1));

Any idea whats wrong with it?

[Fangjun Jiang]

Why need meancell = cellfun(@(T32)mean(T32),T32,'un',0)

Can it be meancell = cellfun(@mean,T32,'un',0)

Answer 6

[Wayne King]

Why didn't you do what I suggested?

detrendcell = cellfun(@(T32)detrend(T32,'linear'),T32,'un',0);
detrendcell = cell2mat(detrendcell);

By the way, 'linear' is the default option for detrend:

so

    detrendcell = cellfun(@detrend,T32,'UniformOutput',false);
    detrendcell = cell2mat(detrendcell);

The above returns a matrix whose columns are your detrended series.

Is the same and a bit cleaner.

[James]

Sorry I was unaware that the default was linear, the other option allowed me to set an optional arguement, thats why I chose it.

I have copy and pasted this exact code, it outputs detrendcell as a single column with 2279310 rows. It detrended each cell and then put them all on top of each other in one single column.

I need each cell detrended and then separately defined.

[Wayne King]

What was the size of your input T32?

[James]

It's a 208x1 cell array with each cell varying from 10957x1 to 10959x1... roughly estimating this 10958*208 = 2279264 so thats why I'm assuming these have all been arranged on top of each other in one column

[Wayne King]

Transpose your input

T32 = T32';

detrendcell = cellfun(@detrend,T32,'UniformOutput',false);

detrendcell = cell2mat(detrendcell);

[James]

This returns the error:

??? Error using ==> cat

CAT arguments dimensions are not consistent.

Error in ==> cell2mat at 77

m{n} = cat(2,c{n,:});

Error in ==> Untitled2 at 28

detrendcell = cell2mat(detrendcell)