逻辑回归与正规化高级优化（MATLAB && Octave version）

一、MEAN FUNCTION:

Data=load('ex2data2.txt');
X=Data(:,[1 2]);
y=Data(:,3);
plotData(X,y);
xlabel('Microchip Test 1');
ylabel('Microchip Test 2');
legend('y=1','y=0');
X28 = mapFeature(X(:,1), X(:,2));
initial_theta = zeros(size(X28,2), 1);
lambda =1;
options= optimset('display','iter','Gradobj','on','maxiter',400,'Algorithm','quasi-newton');
[theta,J,exit_flag]=fminunc(@(t)(costFunctionReg(t, X28, y,lambda)), initial_theta, options);
plotDecisionBoundary(theta, X28, y);
hold on;
title(sprintf('lambda = %g', lambda))

% Labels and Legend
xlabel('Microchip Test 1')
ylabel('Microchip Test 2')

legend('y = 1', 'y = 0', 'Decision boundary')
hold off;

二、FUNCTION HANDLES

（1）BOUNDARY FUNCTION

function plotDecisionBoundary(theta, X, y)
%PLOTDECISIONBOUNDARY Plots the data points X and y into a new figure with
%the decision boundary defined by theta
%   PLOTDECISIONBOUNDARY(theta, X,y) plots the data points with + for the 
%   positive examples and o for the negative examples. X is assumed to be 
%   a either 
%   1) Mx3 matrix, where the first column is an all-ones column for the 
%      intercept.
%   2) MxN, N>3 matrix, where the first column is all-ones

% Plot Data
plotData(X(:,2:3), y);
hold on

if size(X, 2) <= 3
    % Only need 2 points to define a line, so choose two endpoints
    plot_x = [min(X(:,2))-2,  max(X(:,2))+2];

    % Calculate the decision boundary line
    plot_y = (-1./theta(3)).*(theta(2).*plot_x + theta(1));

    % Plot, and adjust axes for better viewing
    plot(plot_x, plot_y,'G','linewidth',1)
    
    % Legend, specific for the exercise
    legend('Admitted', 'Not admitted', 'Decision Boundary')
    axis([30, 100, 30, 100])
else
    % Here is the grid range
    u = linspace(-1, 1.5, 50);
    v = linspace(-1, 1.5, 50);

    z = zeros(length(u), length(v));
    % Evaluate z = theta*x over the grid
    for i = 1:length(u)
        for j = 1:length(v)
            z(i,j) = mapFeature(u(i), v(j))*theta;
        end
    end
    z = z'; % important to transpose z before calling contour

    % Plot z = 0
    % Notice you need to specify the range [0, 0]
    contour(u, v, z, [0, 0],'G', 'LineWidth', 1)
end
hold off

enD

（2）CostFunction && GradientDescent Funciton

function [J, grad] = costFunctionReg(theta, X, y, lambda)

m = length(y); % number of training examples

% You need to return the following variables correctly 
grad = zeros(size(theta));
J=0;

g=sigmoid(X*theta);
h=0;
for i=1:m
    h=h+y(i)*log(g(i))+(1-y(i))*log(1-g(i));
end
J=J-1/m*h;
h=0;
for i=2:size(theta)
    h=h+theta(i)^2;
end
J=J+lambda*h/(2*m);
% Gradient descent
 h=0;
 for i=1:m
     h=h+(g(i)-y(i))*X(i,1);
 end
 grad(1)=h/m;
 for i=2:size(theta)
     h=0;
     for j=1:m
         h=h+(g(j)-y(j))*X(j,i);
     end
 grad(i)=h/m+lambda/m*theta(i);
 end

end
(3)MapFeature Function to return more features(Eg:degree=6)

function out = mapFeature(X1, X2)

degree = 6;
out = ones(size(X1(:,1)));
for i = 1:degree
    for j = 0:i
        out(:, end+1) = (X1.^(i-j)).*(X2.^j);
    end
end

end

(4)Plot Data Function to classify Exmples of the Data [X,y].

function plotData(X, y)

pos=find(y==1);
neg=find(y==0);
plot(X(pos,1),X(pos,2),'k+','linewidth',2);
plot(X(neg,1),X(neg,2),'ko','markersize',7,'markerfacecolor','y');

hold off;

end