Difference between revisions of "FctExampleCode"
(→OLSest.m) |
(→OLSest.m) |
||
Line 71: | Line 71: | ||
bse = sqrt(diag(bse)); | bse = sqrt(diag(bse)); | ||
tstat = b./bse; | tstat = b./bse; | ||
− | |||
ym = y - mean(y); | ym = y - mean(y); | ||
r2 = 1 - (res'*res)/(ym'*ym); | r2 = 1 - (res'*res)/(ym'*ym); | ||
adjr2 = 1 - (n-1)*(1-r2)/(n-k); | adjr2 = 1 - (n-1)*(1-r2)/(n-k); | ||
fstat = ((((ym'*ym))-(res'*res))/(k-1))/((res'*res)/(n-k)); | fstat = ((((ym'*ym))-(res'*res))/(k-1))/((res'*res)/(n-k)); | ||
− | |||
dw = corrcoef([res(1:end-1) res(2:end)]); | dw = corrcoef([res(1:end-1) res(2:end)]); | ||
dw = 2*(1-dw(2,1)); | dw = 2*(1-dw(2,1)); | ||
+ | |||
+ | if output | ||
+ | |||
+ | % calculate p values (requires either MATLAB stats toolbox or NAG toolbox | ||
+ | |||
+ | try % if stats toolbox is available | ||
+ | pval = 2*(1-tcdf(abs(tstat),n-k)); | ||
+ | pvalf = 1- fcdf(fstat,k-1,n-k); | ||
+ | catch | ||
+ | try % if NAG toolbox is available | ||
+ | pval = 2*(1-g01eb(abs(tstat),n-k)); | ||
+ | pvalf = g01ed(fstat,k-1,n-k,'tail','U'); | ||
+ | catch | ||
+ | pval = -999*ones(size(tstat)); | ||
+ | pvalf = -999; | ||
+ | end | ||
+ | end | ||
− | |||
fprintf('===========================================================\n'); | fprintf('===========================================================\n'); | ||
fprintf('===== Regression Output ==================================\n'); | fprintf('===== Regression Output ==================================\n'); | ||
Line 96: | Line 110: | ||
fprintf(' Durbin-Watson = %5.4f\n',dw); | fprintf(' Durbin-Watson = %5.4f\n',dw); | ||
fprintf('===========================================================\n'); | fprintf('===========================================================\n'); | ||
+ | fprintf('== p-values of -999 indicate that neither the stat ========\n'); | ||
+ | fprintf('== nor the NAG toolbox were available =====================\n'); | ||
+ | |||
end | end | ||
− | temp = 2; | + | temp = 2; |
Revision as of 21:23, 4 October 2012
Example MATLAB Code
This code is to be used with the Function discussion.
FunctionExample.m
Copy this code into a m file which you call FunctionExample.m.
% This code loads data from a spreadsheet and uses OLSest to run a
% regresion
[data,titles]=xlsread('OLSexample.xls');
n = size(data,1);
depvar = data(:,1);
expvar = [ones(n,1) data(:,2:end)];
[bhat,bhatse,resids,obs,resss,rsq] = OLSest(depvar,expvar,0);
disp(bhat);
temp = 2;
OLSest.m
Copy this code into a m file which you call OLSest.m or download file from here (OLSest.m).
<source>
function [b,bse,res,n,rss,r2] = OLSest(y,x,output);
% This function performs an OLS estimation
% function [b,bse,res,n,rss,r2] = OLSest(y,x,output)
% input: y, vector with dependent variable
% x, matrix with explanatory variable
% function will automatically add a constant if the first col
% is not a vector of ones
% output, 1 = printed output
% output: b, estimated parameters
% bse, standard errors for bhat
% res, estimated residuals
% n, number of observations used
% rss, residual sum of squares
% r2, Rsquared
% select those rows that have observations for all variables
ninit = length(y);
testnan = [isnan(y) isnan(x)];
testnan = (sum(testnan,2)==0);
y = y(testnan);
x = x(testnan,:);
% test whether first column is vector of ones
temp = (x(x(:,1)==1));
if length(temp) ~= length(x)
x = [ones(length(x),1) x]; % add constant of not included in x
end
[n,k] = size(x); % sample size - n, number of explan vars (incl constant) - k
xxi = inv(x'*x); % Note that this is the inefficient way of calculating
% the inverse of x'*x, but as xxi is required later for
% the calculation of bse, we are not really loosing
% anything
b = xxi*x'*y;
res = y - x*b;
rss = res'*res;
ssq = rss/(n-k);
s = sqrt(ssq);
bse = ssq*xxi;
bse = sqrt(diag(bse));
tstat = b./bse;
ym = y - mean(y);
r2 = 1 - (res'*res)/(ym'*ym);
adjr2 = 1 - (n-1)*(1-r2)/(n-k);
fstat = ((((ym'*ym))-(res'*res))/(k-1))/((res'*res)/(n-k));
dw = corrcoef([res(1:end-1) res(2:end)]);
dw = 2*(1-dw(2,1));
if output
% calculate p values (requires either MATLAB stats toolbox or NAG toolbox
try % if stats toolbox is available
pval = 2*(1-tcdf(abs(tstat),n-k));
pvalf = 1- fcdf(fstat,k-1,n-k);
catch
try % if NAG toolbox is available
pval = 2*(1-g01eb(abs(tstat),n-k));
pvalf = g01ed(fstat,k-1,n-k,'tail','U');
catch
pval = -999*ones(size(tstat));
pvalf = -999;
end
end
fprintf('===========================================================\n');
fprintf('===== Regression Output ==================================\n');
fprintf('Obs used = %4.0f, missing obs = %4.0f \n',n,(ninit-n));
fprintf('Rsquared = %5.4f \n',r2);
fprintf('adj_Rsq = %5.4f \n',adjr2);
fprintf('===== Estimated Model Parameters ==========================\n');
fprintf('= Par se(Par) t(Par) pval ==================\n');
format short;
disp([b bse tstat pval]);
fprintf('===== Model Statistics ====================================\n');
fprintf(' Fstat = %5.4f (%5.4f)\n',[fstat;pvalf]);
fprintf(' standard error = %5.4f\n',sqrt(ssq));
fprintf(' RSS = %5.4f\n',rss);
fprintf(' Durbin-Watson = %5.4f\n',dw);
fprintf('===========================================================\n');
fprintf('== p-values of -999 indicate that neither the stat ========\n');
fprintf('== nor the NAG toolbox were available =====================\n');
end
temp = 2;
Copy this code into a m file which you call FunctionExample.m.
% This code loads data from a spreadsheet and uses OLSest to run a
% regresion
[data,titles]=xlsread('OLSexample.xls');
n = size(data,1);
depvar = data(:,1);
expvar = [ones(n,1) data(:,2:end)];
[bhat,bhatse,resids,obs,resss,rsq] = OLSest(depvar,expvar,0);
disp(bhat);
temp = 2;
OLSest.m
Copy this code into a m file which you call OLSest.m or download file from here (OLSest.m).
<source>
function [b,bse,res,n,rss,r2] = OLSest(y,x,output); % This function performs an OLS estimation % function [b,bse,res,n,rss,r2] = OLSest(y,x,output) % input: y, vector with dependent variable % x, matrix with explanatory variable % function will automatically add a constant if the first col % is not a vector of ones % output, 1 = printed output % output: b, estimated parameters % bse, standard errors for bhat % res, estimated residuals % n, number of observations used % rss, residual sum of squares % r2, Rsquared
% select those rows that have observations for all variables ninit = length(y); testnan = [isnan(y) isnan(x)]; testnan = (sum(testnan,2)==0); y = y(testnan); x = x(testnan,:); % test whether first column is vector of ones temp = (x(x(:,1)==1)); if length(temp) ~= length(x)
x = [ones(length(x),1) x]; % add constant of not included in x
end
[n,k] = size(x); % sample size - n, number of explan vars (incl constant) - k xxi = inv(x'*x); % Note that this is the inefficient way of calculating
% the inverse of x'*x, but as xxi is required later for % the calculation of bse, we are not really loosing % anything
b = xxi*x'*y; res = y - x*b; rss = res'*res; ssq = rss/(n-k); s = sqrt(ssq); bse = ssq*xxi; bse = sqrt(diag(bse)); tstat = b./bse; ym = y - mean(y); r2 = 1 - (res'*res)/(ym'*ym); adjr2 = 1 - (n-1)*(1-r2)/(n-k); fstat = ((((ym'*ym))-(res'*res))/(k-1))/((res'*res)/(n-k)); dw = corrcoef([res(1:end-1) res(2:end)]); dw = 2*(1-dw(2,1));
if output
% calculate p values (requires either MATLAB stats toolbox or NAG toolbox
try % if stats toolbox is available
pval = 2*(1-tcdf(abs(tstat),n-k)); pvalf = 1- fcdf(fstat,k-1,n-k);
catch
try % if NAG toolbox is available pval = 2*(1-g01eb(abs(tstat),n-k)); pvalf = g01ed(fstat,k-1,n-k,'tail','U'); catch pval = -999*ones(size(tstat)); pvalf = -999; end
end
fprintf('===========================================================\n'); fprintf('===== Regression Output ==================================\n'); fprintf('Obs used = %4.0f, missing obs = %4.0f \n',n,(ninit-n)); fprintf('Rsquared = %5.4f \n',r2); fprintf('adj_Rsq = %5.4f \n',adjr2); fprintf('===== Estimated Model Parameters ==========================\n'); fprintf('= Par se(Par) t(Par) pval ==================\n'); format short; disp([b bse tstat pval]); fprintf('===== Model Statistics ====================================\n'); fprintf(' Fstat = %5.4f (%5.4f)\n',[fstat;pvalf]); fprintf(' standard error = %5.4f\n',sqrt(ssq)); fprintf(' RSS = %5.4f\n',rss); fprintf(' Durbin-Watson = %5.4f\n',dw); fprintf('===========================================================\n'); fprintf('== p-values of -999 indicate that neither the stat ========\n'); fprintf('== nor the NAG toolbox were available =====================\n');
end
temp = 2;