added: Ex5 unit tests

ex5/test_ex5.m

@@ -0,0 +1,18 @@
+%!test test_sanity()
+
+% https://class.coursera.org/ml-005/forum/thread?thread_id=2137
+
+% 1 / 5	Regularized Linear Regression Cost Function
+%!test test_linearRegCostFunction_cost()
+
+% 2 / 5	Regularized Linear Regression Gradient
+%!test test_linearRegCostFunction_grad()
+
+% 3 / 5	Learning Curve
+%!test test_learningCurve()
+
+% 4 / 5	Polynomial Feature Mapping
+%!test test_polyFeatures()
+
+% 5 / 5	Cross Validation Curve
+%!test test_validationCurve()

ex5/test_learningCurve.m

@@ -0,0 +1,32 @@
+function test_learningCurve()
+% 3 / 5	Learning Curve
+    epsilon = 1e-3;
+
+    [error_train error_val] = learningCurve([ones(10,1) sec(1:1.5:15)' tan(1:1.5:15)'], cot(1:3:30)', ...
+        [1 23.5 12.4; 1 64.3 10.1; 1 76.4 9.8; 1 34.2 15.2; 1 59.5 13.5], [13;24;53;34;23], 0.1);
+
+    et = [0.00000
+        0.00000
+        0.00003
+        0.03773
+        0.12896
+        0.29892
+        1.59607
+        341.18606
+        331.68731
+        325.76182];
+    ev = [5.0523e+02
+        5.9170e+02
+        7.5200e+02
+        5.3129e+02
+        3.4625e+02
+        7.3199e+02
+        1.6288e+03
+        7.3256e+04
+        4.9892e+04
+        1.12171e+05];
+
+    assert(error_train, et, epsilon);
+    assert(error_val/1000, ev/1000, epsilon);
+
+endfunction

ex5/test_linearRegCostFunction_cost.m

@@ -0,0 +1,16 @@
+function test_linearRegCostFunction_cost()
+% 1 / 5	Regularized Linear Regression Cost Function
+    epsilon = 1e-4;
+
+    X = [1 2; 1 3; 1 4; 1 5; 1 6];
+    y = [7; 6; 4; 3; 2];
+    theta = [1; 1];
+    J = linearRegCostFunction(X, y, theta, 1);
+    assert(J, 5.6, epsilon);
+
+    % https://class.coursera.org/ml-005/forum/thread?thread_id=2137
+    [J] = linearRegCostFunction([ones(10,1) sec(1:1.5:15)' tan(1:1.5:15)'], cot(1:3:30)', ...
+        [0.15 0.25 0.35]', 0.1);
+    assert(J, 644.5304, epsilon);
+
+endfunction

ex5/test_linearRegCostFunction_grad.m

@@ -0,0 +1,14 @@
+function test_linearRegCostFunction_grad()
+% 2 / 5	Regularized Linear Regression Gradient
+    epsilon = 1e-4;
+
+    [J grad] = linearRegCostFunction([1 2; 1 3; 1 4; 1 5; 1 6], [7; 6; 4; 3; 2], [1; 1], 1);
+    assert(J, 5.60, epsilon);
+    assert(grad, [0.60; 7.20], epsilon);
+
+    [J grad] = linearRegCostFunction([1 23.5 12.4; 1 64.3 10.1; 1 76.4 9.8; 1 34.2 15.2; 1 59.5 13.5], ...
+        [13;24;53;34;23], [3.2;2.3;1.2], 1.2);
+    assert(J, 6396.44119, epsilon);
+    assert(grad, [107.074; 6210.2529; 1258.7064], epsilon);
+
+endfunction

ex5/test_polyFeatures.m

@@ -0,0 +1,10 @@
+function test_polyFeatures()
+% 4 / 5	Polynomial Feature Mapping
+    epsilon = 1e-3;
+    Xpoly = polyFeatures([2; 5; 7], 4);
+    a = [2      4      8     16
+        5     25    125    625
+        7     49    343   2401];
+    assert(Xpoly, a, epsilon);
+
+endfunction

ex5/test_sanity.m

@@ -0,0 +1,4 @@
+function test_sanity ()
+  % make sure tests are running correctly
+  assert(1,1);
+endfunction

ex5/test_validationCurve.m

@@ -0,0 +1,43 @@
+function test_validationCurve()
+% 5 / 5	Cross Validation Curve
+    epsilon = 1e-3;
+
+    [lambda_vec error_train error_val] = validationCurve([ones(10,1) sec(1:1.5:15)' tan(1:1.5:15)'], cot(1:3:30)', ...
+        [1 23.5 12.4; 1 64.3 10.1; 1 76.4 9.8; 1 34.2 15.2; 1 59.5 13.5], [13;24;53;34;23]);
+
+    lv = [0.00000
+        0.00100
+        0.00300
+        0.01000
+        0.03000
+        0.10000
+        0.30000
+        1.00000
+        3.00000
+        10.00000];
+    etr = [325.751
+        325.751
+        325.751
+        325.751
+        325.752
+        325.762
+        325.847
+        326.723
+        332.616
+        364.812];
+    ecv = [1.1340e+05
+        1.1339e+05
+        1.1336e+05
+        1.1327e+05
+        1.1303e+05
+        1.1217e+05
+        1.0978e+05
+        1.0202e+05
+        8.4041e+04
+        4.8477e+04];
+
+    assert(lambda_vec, lv, epsilon);
+    assert(error_train, etr, epsilon);
+    assert(error_val/10000, ecv/10000, epsilon);
+
+endfunction