import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.cross_validation import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.datasets import load_breast_cancer
import matplotlib.pyplot as plt

data = load_breast_cancer()
np.random.seed(0)
X = data.data
Y = data.target
X = X / X.max(axis=0)
X_train, X_test, Y_train, Y_test = train_test_split(X, Y.ravel(), test_size=0.33)


class LogReg(object):
    def __init__(self):
        pass

    def sigmoid(self, X):
        return 1 / (1 + np.exp(-X))

    def hypothesis(self, Xp):
        return Xp.dot(self.theta)

    def cost(self):
        r = (self.hypothesis(self.x))
        r=r/r.max(axis=0)
        r=np.absolute(r)
        p1=0
        p2=0
        for oy in range(len(self.y)):
            if self.y[oy]==1:
                p1+=float(np.log(r[oy]))
            else:
                p2+=float(np.log(1-np.log(r[oy])))
        r=-(p1+p2)/len(self.y)
        return r
    def fit(self, X, Y, alpha, itr):
        self.x = np.array(X)
        self.y = np.array(Y).reshape(-1, 1)
        self.alpha = alpha
        self.cst = []
        self.theta = np.zeros([self.x.shape[1], self.y.shape[1]])
        for i in range(itr):
            self.error = self.hypothesis(self.x) - self.y
            self.theta = self.theta - (self.alpha / len(self.x)) * self.x.T.dot(self.error)
            self.cst.append(self.cost())

    def costgraph(self):
        plt.plot(range(len(self.cst)), self.cst)
        plt.grid(True)
        plt.show()

    def preditc(self, X):
        output = self.hypothesis(X)
        output[output>=0.5]=1
        output[output <0.5] = 0
        return output


a = LogReg()
a.fit(X_train, Y_train, alpha=.07, itr=5000)
output = a.preditc(X_test)

print("Scratch Accuracy : ",accuracy_score(Y_test, output))


model = LogisticRegression()
model.fit(X_train, Y_train)

print("Scikit Accuracy (With Default Parameters: ",accuracy_score(Y_test, model.predict(X_test)))

a.costgraph()