hide long namesshow long names
hide short namesshow short names
Integer type:  int32  int64  nag_int  show int32  show int32  show int64  show int64  show nag_int  show nag_int
PDF version (NAG web site, 64-bit version, 64-bit version)
Chapter Contents
Chapter Introduction
NAG Toolbox

NAG Toolbox: nag_correg_quantile_linreg_easy (g02qf)

▸▿ Contents

    1  Purpose
    2  Syntax
    3  Description
    4  References
▸▿ 5  Parameters
    6  Error Indicators and Warnings
    7  Accuracy
    8  Further Comments
    9  Example

Purpose

nag_correg_quantile_linreg_easy (g02qf) performs a multiple linear quantile regression, returning the parameter estimates and associated confidence limits based on an assumption of Normal, independent, identically distributed errors. nag_correg_quantile_linreg_easy (g02qf) is a simplified version of nag_correg_quantile_linreg (g02qg).

Syntax

[df, b, bl, bu, info, ifail] = g02qf(x, y, tau, 'n', n, 'm', m, 'ntau', ntau)
[df, b, bl, bu, info, ifail] = nag_correg_quantile_linreg_easy(x, y, tau, 'n', n, 'm', m, 'ntau', ntau)

Description

Given a vector of n observed values, y = y i : i = 1, 2, …, n , an n×p design matrix X, a column vector, x, of length p holding the ith row of X and a quantile τ ∈ 0 , 1 , nag_correg_quantile_linreg_easy (g02qf) estimates the p-element vector β as the solution to
minimize β ∈ ℝ p ∑ i=1 n ρ τ y i - xiT β (1)
where ρ τ  is the piecewise linear loss function ρ τ z = z ⁢ τ - I z < 0 , and I z < 0  is an indicator function taking the value 1 if z<0 and 0 otherwise.
nag_correg_quantile_linreg_easy (g02qf) assumes Normal, independent, identically distributed (IID) errors and calculates the asymptotic covariance matrix from
Σ = τ 1 - τ n s τ 2 XT X -1  
where s is the sparsity function, which is estimated from the residuals, ri = yi - xiT ⁢ β^  (see Koenker (2005)).
Given an estimate of the covariance matrix, Σ^, lower, β^L, and upper, β^U, limits for a 95% confidence interval are calculated for each of the p parameters, via
β^ Li = β^ i - t n-p , 0.975 Σ^ ii , β^ Ui = β^ i + t n-p , 0.975 Σ^ ii  
where tn-p,0.975 is the 97.5 percentile of the Student's t distribution with n-k degrees of freedom, where k is the rank of the cross-product matrix XTX.
Further details of the algorithms used by nag_correg_quantile_linreg_easy (g02qf) can be found in the documentation for nag_correg_quantile_linreg (g02qg).

References

Koenker R (2005) Quantile Regression Econometric Society Monographs, Cambridge University Press, New York

Parameters

Compulsory Input Parameters

1:     xnm – double array
X, the design matrix, with the ith value for the jth variate supplied in xij, for i=1,2,…,n and j=1,2,…,m.
2:     yn – double array
y, the observations on the dependent variable.
3:     tauntau – double array
The vector of quantiles of interest. A separate model is fitted to each quantile.
Constraint: ε<taul<1-ε where ε is the machine precision returned by nag_machine_precision (x02aj), for l=1,2,…,ntau.

Optional Input Parameters

1:     n – int64int32nag_int scalar
Default: the dimension of the array y and the first dimension of the array x. (An error is raised if these dimensions are not equal.)
n, the number of observations in the dataset.
Constraint: n≥2.
2:     m – int64int32nag_int scalar
Default: the second dimension of the array x.
p, the number of variates in the model.
Constraint: 1≤m<n.
3:     ntau – int64int32nag_int scalar
Default: the dimension of the array tau.
The number of quantiles of interest.
Constraint: ntau≥1.

Output Parameters

1:     df – double scalar
The degrees of freedom given by n-k, where n is the number of observations and k is the rank of the cross-product matrix XTX.
2:     bmntau – double array
β^, the estimates of the parameters of the regression model, with bjl containing the coefficient for the variable in column j of x, estimated for τ=taul.
3:     blmntau – double array
β^L, the lower limit of a 95% confidence interval for β^, with bljl holding the lower limit associated with bjl.
4:     bumntau – double array
β^U, the upper limit of a 95% confidence interval for β^, with bujl holding the upper limit associated with bjl.
5:     infontau – int64int32nag_int array
infol holds additional information concerning the model fitting and confidence limit calculations when τ=taul.
Code Warning
0 Model fitted and confidence limits calculated successfully.
1 The function did not converge whilst calculating the parameter estimates. The returned values are based on the estimate at the last iteration.
2 A singular matrix was encountered during the optimization. The model was not fitted for this value of τ.
8 The function did not converge whilst calculating the confidence limits. The returned limits are based on the estimate at the last iteration.
16 Confidence limits for this value of τ could not be calculated. The returned upper and lower limits are set to a large positive and large negative value respectively.
It is possible for multiple warnings to be applicable to a single model. In these cases the value returned in info is the sum of the corresponding individual nonzero warning codes.
6:     ifail – int64int32nag_int scalar
ifail=0 unless the function detects an error (see Error Indicators and Warnings).

Error Indicators and Warnings

Errors or warnings detected by the function:
   ifail=11
Constraint: n≥2.
   ifail=21
Constraint: 1≤m<n.
   ifail=51
Constraint: ntau≥1.
   ifail=61
On entry is invalid.
   ifail=111
A potential problem occurred whilst fitting the model(s).
Additional information has been returned in info.
   ifail=-99
An unexpected error has been triggered by this routine. Please contact NAG.
   ifail=-399
Your licence key may have expired or may not have been installed correctly.
   ifail=-999
Dynamic memory allocation failed.

Accuracy

Not applicable.

Further Comments

Calling nag_correg_quantile_linreg_easy (g02qf) is equivalent to calling nag_correg_quantile_linreg (g02qg) with

Example

A quantile regression model is fitted to Engels 1857 study of household expenditure on food. The model regresses the dependent variable, household food expenditure, against household income. An intercept is included in the model by augmenting the dataset with a column of ones.

Open in the MATLAB editor: g02qf_example

function g02qf_example


fprintf('g02qf example results\n\n');

n = 235;
x = zeros(n,2);
x(:,1) = ones(n,1);
x(:,2) = ...
    [ 420.1577  800.7990  643.3571  541.4117 1245.6964 2551.6615  901.1575 ...
     1201.0002 1795.3226  639.0802  634.4002 1165.7734  750.8756  956.2315 ...
      815.6212  945.7989 1148.6010 1264.2066  829.3979 1768.8236 1095.4056 ...
      979.1648 2822.5330  447.4479 1309.8789  922.3548 1178.9742 1492.3987 ...
     2293.1920  975.8023  502.8390  627.4726 1017.8522  616.7168  889.9809 ...
      423.8798  790.9225 1162.2000  558.7767  555.8786 1197.0794  943.2487 ...
      713.4412  530.7972 1348.3002  838.7561 1142.1526 2340.6174  535.0766 ...
     1088.0039  587.1792  596.4408  484.6612 1540.9741  924.5619 1536.0201 ...
     1115.8481  487.7583  678.8974 1044.6843  692.6397  671.8802 1389.7929 ...
      997.8770  690.4683 2497.7860  506.9995  860.6948 1585.3809  654.1587 ...
      873.3095 1862.0438  933.9193  894.4598 2008.8546  433.6813 1148.6470 ...
      697.3099  587.5962  926.8762  571.2517  896.4746  839.0414  598.3465 ...
      454.4782  829.4974  461.0977  584.9989 1264.0043  977.1107  800.7990 ...
     1937.9771  883.9849  502.4369  698.8317  718.3594  713.5197  920.4199 ...
      543.8971  906.0006 1897.5711 1587.3480  880.5969  891.6824 4957.8130 ...
      796.8289  889.6784  969.6838  854.8791 1221.4818  419.9980 1167.3716 ...
      544.5991  561.9990  523.8000 1031.4491  689.5988  670.7792 1462.9497 ...
     1398.5203  377.0584  830.4353  820.8168  851.5430  975.0415  875.1716 ...
     1121.0937 1337.9983 1392.4499  625.5179  867.6427 1256.3174  805.5377 ...
      725.7459 1362.8590  558.5812  989.0056 1999.2552  884.4005 1525.0005 ...
     1209.4730 1257.4989  672.1960 1125.0356 2051.1789  923.3977 1827.4010 ...
     1466.3330  472.3215 1014.1540  730.0989  590.7601  880.3944 2432.3910 ...
      831.7983  873.7375  940.9218 1139.4945  951.4432 1177.8547  507.5169 ...
      473.0022 1222.5939  576.1972  601.0030 1519.5811  696.5991  713.9979 ...
      687.6638  650.8180  829.2984  953.1192  949.5802  959.7953  953.1192 ...
      497.1193 1212.9613  953.1192  570.1674  958.8743  939.0418  724.7306 ...
     1129.4431 1283.4025  408.3399 1943.0419 1511.5789  638.6713  539.6388 ...
     1342.5821 1225.7890  463.5990  511.7980  715.3701  562.6400  689.7988 ...
      800.4708  736.7584 1532.3074  975.5974 1415.4461 1056.0808 1613.7565 ...
     2208.7897  387.3195  608.5019  636.0009  387.3195  958.6634  759.4010 ...
      410.9987  835.9426 1078.8382  499.7510 1024.8177  748.6413  832.7554 ...
     1006.4353  987.6417  614.9986  726.0000  788.0961  887.4658  494.4174 ...
     1020.0225 1595.1611  776.5958 1230.9235 1807.9520  415.4407  440.5174 ...
      541.2006  581.3599  743.0772 1057.6767];

y = [ 255.8394  572.0807  459.8177  310.9587  907.3969  863.9199  485.6800 ...
      811.5776  831.4407  402.9974  427.7975  534.7610  495.5608  649.9985 ...
      392.0502  633.7978  860.6002  934.9752  630.7566 1143.4211  813.3081 ...
      700.4409 2032.6792  263.7100  830.9586  590.6183  769.0838  815.3602 ...
     1570.3911  630.5863  338.0014  483.4800  645.9874  412.3613  600.4804 ...
      319.5584  520.0006  696.2021  348.4518  452.4015  774.7962  614.5068 ...
      512.7201  390.5984  662.0096  658.8395  612.5619 1504.3708  392.5995 ...
      708.7622  406.2180  443.5586  296.9192  692.1689  640.1164 1071.4627 ...
      588.1371  333.8394  496.5976  511.2609  466.9583  503.3974  700.5600 ...
      543.3969  357.6411 1301.1451  317.7198  430.3376  879.0660  424.3209 ...
      624.6990  912.8851  518.9617  582.5413 1509.7812  338.0014  580.2215 ...
      484.0605  419.6412  543.8807  399.6703  476.3200  588.6372  444.1001 ...
      386.3602  627.9999  248.8101  423.2783  712.1012  527.8014  503.3572 ...
      968.3949  500.6313  354.6389  482.5816  436.8107  497.3182  593.1694 ...
      374.7990  588.5195 1033.5658  726.3921  654.5971  693.6795 1827.2000 ...
      550.7274  693.6795  523.4911  528.3770  761.2791  334.9998  640.4813 ...
      361.3981  473.2009  401.3204  628.4522  581.2029  435.9990  771.4486 ...
      929.7540  276.5606  757.1187  591.1974  588.3488  821.5970  637.5483 ...
      664.1978 1022.3202  674.9509  444.8602  679.4407  776.7589  462.8995 ...
      538.7491  959.5170  377.7792  679.9981 1250.9643  553.1504  977.0033 ...
      737.8201  810.8962  561.2015  810.6772 1067.9541  728.3997  983.0009 ...
     1049.8788  372.3186  708.8968  522.7012  361.5210  633.1200 1424.8047 ...
      620.8006  631.7982  517.9196  819.9964  608.6419  830.9586  360.8780 ...
      300.9999  925.5795  395.7608  377.9984 1162.0024  442.0001  397.0015 ...
      383.4580  404.0384  588.5195  621.1173  670.7993  681.7616  621.1173 ...
      297.5702  807.3603  621.1173  353.4882  696.8011  548.6002  383.9376 ...
      811.1962  745.2353  284.8008 1305.7201  837.8005  431.1000  442.0001 ...
      795.3402  801.3518  353.6013  418.5976  448.4513  468.0008  508.7974 ...
      577.9111  526.7573  883.2780  570.5210  890.2390  742.5276  865.3205 ...
     1318.8033  242.3202  444.5578  331.0005  242.3202  680.4198  416.4015 ...
      266.0010  576.2779  596.8406  408.4992  708.4787  429.0399  614.7588 ...
      734.2356  619.6408  385.3184  433.0010  400.7990  515.6200  327.4188 ...
      775.0209 1138.1620  485.5198  772.7611  993.9630  305.4390  306.5191 ...
      299.1993  468.0008  522.6019  750.3202];

tau = [0.10; 0.25; 0.50; 0.75; 0.90];

% Call the model fitting routine
[df, b, bl, bu, info, ifail] = g02qf( ...
                                      x, y, tau);

% Display the parameter estimates
% plot setup
t = '\tau';
fig1 = figure;
hold on;
plot(x(:,2),y,'+r');
tt{1} = 'data';
% loop over tau
for l=1:numel(tau)
  fprintf('\nQuantile: %6.3f\n\n', tau(l));
  fprintf('        Lower   Parameter   Upper\n');
  fprintf('        Limit   Estimate    Limit\n');
  for j=1:2
    fprintf('%3d   %7.3f   %7.3f   %7.3f\n', j, bl(j,l), b(j,l), bu(j,l));
  end
  fprintf('\n');
  plot([0 (2000-b(1,l))/b(2,l)],[b(1,l) 2000]);
  tt{l+1} = sprintf('%s = %4.2f',t,tau(l));
end
% plot labels
legend(tt,'Location','SouthEast')
xlabel('Household income');
ylabel('Household food expenditure');
title({'Quantile Regression', ...
       ' Study of Household Expenditure on Food', ...
       'Engels 1857'});
axis([0 5000 0 2000]);
hold off;


g02qf example results


Quantile:  0.100

        Lower   Parameter   Upper
        Limit   Estimate    Limit
  1    74.946   110.142   145.337
  2     0.370     0.402     0.433


Quantile:  0.250

        Lower   Parameter   Upper
        Limit   Estimate    Limit
  1    64.232    95.483   126.735
  2     0.446     0.474     0.502


Quantile:  0.500

        Lower   Parameter   Upper
        Limit   Estimate    Limit
  1    55.399    81.482   107.566
  2     0.537     0.560     0.584


Quantile:  0.750

        Lower   Parameter   Upper
        Limit   Estimate    Limit
  1    41.372    62.396    83.421
  2     0.625     0.644     0.663


Quantile:  0.900

        Lower   Parameter   Upper
        Limit   Estimate    Limit
  1    26.829    67.351   107.873
  2     0.650     0.686     0.723
g02qf_fig1.png
PDF version (NAG web site, 64-bit version, 64-bit version)
Chapter Contents
Chapter Introduction
NAG Toolbox
© The Numerical Algorithms Group Ltd, Oxford, UK. 2009–2015