# coding=utf-8 
 
# 首先利用sklearn的庫(kù)進(jìn)行knn算法的建立與預(yù)測(cè) 
# from sklearn import neighbors 
# from sklearn import datasets 
# 
# knn = neighbors.KNeighborsClassifier()   # 調(diào)用分類器賦在變量knn上 
# 
# iris = datasets.load_iris()   # 返回一個(gè)數(shù)據(jù)庫(kù)，賦值在iris上 
# 
# print iris   # 顯示這個(gè)數(shù)據(jù)集 
# 
# knn.fit(iris.data, iris.target) # fit的第一個(gè)參數(shù) 是特征值矩陣，第二個(gè)參數(shù)是一維的向量 
# 
# predictedLabel = knn.predict([[0.1,0.2,0.3,0.4]]) 
# 
# print predictedLabel 
 
# 下面自己寫(xiě)一個(gè)程序?qū)崿F(xiàn)knn算法 
 
import csv 
import random 
import math 
import operator 
 
# filename是指文件名,split是某一個(gè)數(shù)字，數(shù)字前的數(shù)據(jù)當(dāng)做訓(xùn)練集，數(shù)字后的數(shù)據(jù)當(dāng)做測(cè)試集 
# trainingSet是訓(xùn)練集，testSet是測(cè)試集 
# 函數(shù)作用，加載文件，并將文件通過(guò)隨機(jī)數(shù)的方法分為訓(xùn)練集和測(cè)試集 
def loadDataset(filename, split, trainingSet=[], testSet=[]): 
  with open(filename, 'rb') as csvfile:  # 導(dǎo)入文件為csvfile格式 
    lines = csv.reader(csvfile)   # 讀取所有的行 reader函數(shù)的作用 
    dataset = list(lines)    # 將所有的行轉(zhuǎn)換為list的數(shù)據(jù)節(jié)后 
    for x in range(len(dataset)-1):   # x在總共的行數(shù)中遍歷 
      for y in range(4): 
        dataset[x][y] = float(dataset[x][y]) 
      if random.random() < split: 
        trainingSet.append(dataset[x]) 
      else: 
        testSet.append(dataset[x]) 
 
 
# 函數(shù)作用：計(jì)算歐氏距離 
# 函數(shù)的輸入是兩個(gè)實(shí)例和他們的維度 
def euclideanDistance(instance1, instance2, length): 
  distance = 0 
  for x in range(length):   # 對(duì)于每一個(gè)維度內(nèi)進(jìn)行一個(gè)差的計(jì)算，計(jì)算出所有維度的平方和 
    distance += pow((instance1[x] - instance2[x]),2) 
  return math.sqrt(distance) 
 
# 函數(shù)作用：返回最近的k的neightbor 
# 也就是返回在trainingSet中距離testInstance最近的k個(gè)鄰居 
def getNeigthbors(trainingSet, testInstance, k): 
  distances =[] # 距離的容器，用來(lái)存放所有的距離值 
  length = len(testInstance) - 1 # 用來(lái)存放testInstance的維度 
  for x in range(len(trainingSet)): 
    # 對(duì)于每一個(gè)x 計(jì)算訓(xùn)練集中的數(shù)據(jù)與實(shí)例的距離 
    dist = euclideanDistance(testInstance,trainingSet[x],length) 
    distances.append((trainingSet[x],dist)) 
  # 把這些距離從小到大排起來(lái) 
  distances.sort(key=operator.itemgetter(1)) 
  neighbors = [] 
  for x in range(k): 
    neighbors.append(distances[x][0]) 
  return neighbors    # 返回最近的鄰居 
 
def getResponse(neighbors): 
  classVotes = {} 
  for x in range(len(neighbors)): 
    response = neighbors[x][-1] 
    if response in classVotes: 
      classVotes[response] += 1 
    else: 
      classVotes[response] = 1 
  sortedVotes = sorted(classVotes.iteritems(),key=operator.itemgetter(1),reverse=True) 
  return sortedVotes[0][0] 
 
# 用來(lái)檢驗(yàn)預(yù)測(cè)結(jié)果的正確率 
def getAccuracy(testSet,predictions): 
  correct = 0 
  for x in range(len(testSet)): 
    if testSet[x][-1] == predictions[x]:    # [-1]值的是最后一個(gè)值，也就是每行的最后的值，即為花的分類 
      correct += 1 
  return (correct/float(len(testSet))) * 100.00 
 
 
def main(): 
  # prepare data 
  trainingSet = [] 
  testSet = [] 
  split = 0.67 
  loadDataset('irisdata.txt',split,trainingSet,testSet) # r的作用是防止錯(cuò)誤字符串意思 
  print 'Train Set' + repr(len(trainingSet)) 
  print 'Test Set' + repr(len(testSet)) 
 
  # generate predicitions 
  predicitions = [] 
  k = 3 
  for x in range(len(testSet)): 
    neighbors = getNeigthbors(trainingSet,testSet[x],k) 
    result = getResponse(neighbors) 
    predicitions.append(result) 
    print('> predicition = ' + repr(result) + ', actual = ' +repr(testSet[x][-1])) 
  accuracy = getAccuracy(testSet,predicitions) 
  print('Accuracy:' + repr(accuracy) + '%') 
 
main()