PROG9

 import matplotlib.pyplot as plt

import pandas as pd

import numpy as np

def kernel(point, xmat, k):

    m, n = np.shape(xmat)

    weights = np.mat(np.eye((m)))  # eye - identity matrix

    for j in range(m):

        diff = point - X[j]

        weights[j, j] = np.exp(diff * diff.T / (-2.0 * k**2))

    return weights

def localWeight(point, xmat, ymat, k):

    wei = kernel(point, xmat, k)

    W = (X.T * (wei * X)).I * (X.T * (wei * ymat.T))

    return W

def localWeightRegression(xmat, ymat, k):

    m, n = np.shape(xmat)

    ypred = np.zeros(m)

    for i in range(m):

        ypred[i] = xmat[i] * localWeight(xmat[i], xmat, ymat, k)

    return ypred

def graphPlot(X, ypred):

    sortindex = X[:, 1].argsort(0)  # argsort - index of the smallest

    xsort = X[sortindex][:, 0]

    fig = plt.figure()

    ax = fig.add_subplot(1, 1, 1)

    ax.scatter(bill, tip, color='green')

    ax.plot(xsort[:, 1], ypred[sortindex], color='red', linewidth=5)

    plt.xlabel('Total bill')

    plt.ylabel('Tip')

    plt.show()

data = pd.read_csv('10_tips.csv')

bill = np.array(data.total_bill)

tip = np.array(data.tip)

mbill = np.mat(bill)  # .mat will convert nd array to 2D array

mtip = np.mat(tip)

m = np.shape(mbill)[1]

one = np.mat(np.ones(m))

X = np.hstack((one.T, mbill.T))  # 244 rows, 2 cols

ypred = localWeightRegression(X, mtip, 5)

graphPlot(X, ypred)