How to use cellfun?

Hi all,

I have two 21 x1 cells (t_sq_mtw / t_sq_dot_1) containg 100x18 tables.

The below code calculates three parameters (a0_longhand/ a1_longhand/rSq_longahnd) for one variable (P_acc_z_meancycle) from correspoding first pair of tables.

I was wondering how could I apply cellfun to apply the code to all tables (not just one as in the code below)? Can you help please?

Pref = (t_sq_mtw{1, 1}.P_acc_z_meancycle); %variable one from data set 1
Pa = (t_sq_dot_1{1, 1}.P_acc_z_meancycle); %%variable one from data set 2

    for i=1:length(Pref)
        temp_toprow(i)=(Pref(i,1)-mean(Pref))*(Pa(i,1)-mean(Pa));
        temp_bottomrow(i)=(Pref(i,1)-mean(Pref))^2;
    end

    toprow=sum(temp_toprow);
    bottomrow=sum(temp_bottomrow);
    a1_longhand=toprow/bottomrow;
    a0_longhand=mean(Pa)-(a1_longhand*mean(Pref));
    
    n=length(Pa);
    x=Pref;
    y=Pa;
    toprow=(n*(sum(y.*x)))-(sum(x)*sum(y));
    bottomrow=sqrt(((n*sum(x.^2))-(sum(x))^2)*((n*sum(y.^2))-(sum(y))^2));
    r=toprow/bottomrow;
    rSq_longhand=r^2;
    
    if a1~=a1_longhand
        warning(1)=1;
    else
        warning(1)=0;
    end
    if a0~=a0_longhand
        warning(2)=1;
    else
        warning(2)=0;
    end
    if rSq~=rSq_longhand
        warning(3)=1;
    else
        warning(3)=0;
    end

This is the first part of the code were cellfun is sucessfully used.

% do polyfit() of P_acc_z_meancycle for each pair of tables
    p = cellfun(@(x,y)polyfit(x.P_acc_z_meancycle,y.P_acc_z_meancycle,1), ...
        t_sq_mtw,t_sq_dot_1,'UniformOutput',false);
    p = cell2mat(p); % p is a 21-by-2 matrix of coefficients from polyfit()
    a1 = p(:,1);
    a0 = p(:,2);

% do corrcoef() of P_acc_z_meancycle for each pair of tables:
    r = cellfun(@(x,y)corrcoef(x.P_acc_z_meancycle,y.P_acc_z_meancycle), ...
        t_sq_mtw,t_sq_dot_1,'UniformOutput',false);

    rSq = cellfun(@(x)x(1,2)^2,r); % rSq is a 21-by-1 column vector of r-squared values from corrcoef()

Answer 1

[Voss]

You can make that code into its own function and then use cellfun() to call that function on all pairs of tables:

load('Data1.mat'); % t_sq_mtw
load('Data2.mat'); % t_sq_dot_1

% call get_longhand() on each pair of tables' P_acc_z_meancycle
[a1_longhand,a0_longhand,rSq_longhand] = cellfun( ...
    @(x,y)get_longhand(x.P_acc_z_meancycle,y.P_acc_z_meancycle), ...
    t_sq_mtw,t_sq_dot_1);
disp([a1_longhand a0_longhand rSq_longhand]);

    0.9867    0.1336    0.9849
    1.0177    0.1576    0.9866
    0.9638    0.1300    0.8919
   -0.4362    0.0891    0.3034
    0.7685    0.1178    0.8005
    0.5379    0.1677    0.2836
    1.0999    0.1258    0.9708
    0.9652    0.1466    0.8687
    0.9787    0.2405    0.9272
    1.2832    0.4761    0.6527
    0.9320    0.1208    0.9839
    0.9377    0.0799    0.8746
    0.9965    0.1056    0.9026
    0.0039    0.1945    0.0000
    1.0158    0.1026    0.9746
    1.0483    0.1801    0.9568
    1.2828    0.2525    0.4985
    0.9924    0.1191    0.9674
    0.9760    0.1164    0.9577
    0.5547    0.0885    0.3260
    1.0794    0.2551    0.7127

function [a1_longhand,a0_longhand,rSq_longhand] = get_longhand(Pref,Pa)
    
    for i=1:length(Pref)
        temp_toprow(i)=(Pref(i,1)-mean(Pref))*(Pa(i,1)-mean(Pa));
        temp_bottomrow(i)=(Pref(i,1)-mean(Pref))^2;
    end
    
    toprow=sum(temp_toprow);
    bottomrow=sum(temp_bottomrow);
    a1_longhand=toprow/bottomrow;
    a0_longhand=mean(Pa)-(a1_longhand*mean(Pref));
    
    n=length(Pa);
    x=Pref;
    y=Pa;
    toprow=(n*(sum(y.*x)))-(sum(x)*sum(y));
    bottomrow=sqrt(((n*sum(x.^2))-(sum(x))^2)*((n*sum(y.^2))-(sum(y))^2));
    r=toprow/bottomrow;
    rSq_longhand=r^2;
    
end

[Tomaszzz]

Many thanks!

[Voss]

You're welcome!