solves advoce for applied machine learning quiz

Author: anishLearnsToCode

README.md

@@ -74,3 +74,8 @@ __Instructor__: Andrew Ng.
     - [Neural Network Gradient (Backpropagation)](week5/ex4/checkNNGradients.m) 
     - [Neural Network Cost Function](week5/ex4/nnCostFunction.m)
     - [Regularized Gradient](week5/ex4/checkNNGradients.m)
+
+## Week 6
+### Quizzes
+
+### Programming Exercises

test.m

@@ -1,16 +1,46 @@
 clc;
 clear;
 
-y = [1 1 ; 1 2 ; 1 3];
-y = y == 2;
-disp(y);
-
-function row = labelVector(len, label)
-    row = zeros(len, 1);
-    row(label) = 1;
- endfunction
- 
- disp(labelVector(4, 2)');
- disp(labelVector(3, 1)');
- disp(labelVector(10, 6)');
- 
+function error = logisticMisclassificationError(hypothesis, y, threshold)
+  prediction = hypothesis >= threshold;
+  error = sum(prediction != y) / length(y);
+endfunction
+
+function J = regressionCost(hypothesis, data, result)
+  trainingSampleSize = size(data, 1);
+  J = (1 / (2 * trainingSampleSize)) * sum((data * hypothesis - result) .^ 2);
+endfunction
+
+function grad = regressionGradient(hypothesis, X, y)
+  trainingSampleSize = size(X, 1);
+  grad = (1 / trainingSampleSize) * (((X * hypothesis) - y)' * X)';
+endfunction
+
+function [minCost, hypothesis, costMemory] = gradientDescent(hypothesis, X, y, iterations, learningRate)
+  costMemory = [];
+  for i = 1:iterations
+    hypothesis = hypothesis - learningRate * regressionGradient(hypothesis, X, y);
+    minCost = regressionCost(hypothesis, X, y);
+    costMemory = [costMemory minCost];
+  endfor
+endfunction
+
+hypothesis = [0.3 ; 0.5 ; 0.3 ; 0.9];
+result = [1 ; 0 ; 1 ; 1];
+error = logisticMisclassificationError(hypothesis, result, 0.5);
+disp(error);
+
+hypothesis = [0 ; 3];
+data = [1 1 ; 1 2 ; 1 3];
+y = [1 ; 2 ; 3];
+disp(regressionCost(hypothesis, data, y, 3));
+
+initialTheta = [0 ; 0 ];
+[minCost, hypothesis, costMemory] = gradientDescent(initialTheta, data, y, 100, 0.09);
+disp('min cost');
+disp(minCost);
+
+disp('hypothesis');
+disp(hypothesis);
+
+plot(costMemory);

week5/ex4/nnCostFunction.m

@@ -92,7 +92,6 @@
   Theta2_grad /= m; 
   Theta1_grad /= m; 
 
-
   % Part 3: Implement regularization with the cost function and gradients.
 
   regularizationMaskTheta1 = (lambda / m) * ones(size(Theta1));

week6/advice-for-applying-machine-learning.md

@@ -0,0 +1,11 @@
+# Advice for Applying Machine Learning Quiz
+
+![Question 1](assets/quiz-1.PNG)
+![Question 1](assets/question-1-2.PNG)
+![Question 2](assets/question-2-2.PNG)
+![Question 3](assets/question-3.PNG)
+![Question 3](assets/question-3-2.PNG)
+![Question 4](assets/question-4.PNG)
+![Question 4](assets/question-4-2.PNG)
+![Question 5](assets/question-5.PNG)
+![Question 5](assets/question-5-2.PNG)