# HG changeset patch
# User Jordi Gutiérrez Hermoso <jordigh@octave.org>
# Date 1322427494 18000
# Node ID eddd33e57f6a55ef11fe19305a5104e950d1f96f
# Parent  4b6c21158ba68bd8ed4f5af1cec7743339be6eef
Justify loop in trainLinearReg

diff --git a/learningCurve.m b/learningCurve.m
--- a/learningCurve.m
+++ b/learningCurve.m
@@ -19,7 +19,12 @@
   ## Initialise outputs
   error_train = zeros(m, 1);
   error_val   = zeros(m, 1);
-  
+
+  ## It is not worth getting rid of this loop because the complexity is
+  ## inside trainLinearReg which in turn is inside fmincg. While you
+  ## *could* do some matrix gymnastics to perform a single minimisation
+  ## in a much higher dimensional space instead of m minimimsations, the
+  ## effort is unlikely to produce faster code.
   for i = 1:m
     Xtrain = X(1:i, :);
     ytrain = y(1:i);