第二课:学习knn算法,
简单易于理解:这个算法的重点在于k的取值,是这样的,当你想判别一个未知的点的类别时,我们采取选取离他最近的k个点,这k个点为已知类别,这k中哪一类的数量多,我们就把未知点归为哪一类,简单点就是,周围是什么人,它就是什么人。
这个是一个手写的knn算法,也可以直接调用sklearn库中的函数
from sklearn import neighbors
from sklearn import datasets
knn = neighbors.KNeighborsClassifier()
# 花的萼长,宽,花瓣的长,宽
# 字典
iris = datasets.load_iris()
print(iris)
knn.fit(iris.data, iris.target)
predictedLabel = knn.predict([[6.2, 3.4, 5.4, 2.3]])
print(predictedLabel)
接下来就是自己写的knn算法,不难的
第一步:读取数据,分为测试数据和训练数据集
# 读取文件数据--csv
# 随机加进trainset和testset
def loadDataset(filename, split, trainSet=[], testSet=[]):
#这里是一个以csv读取txt,并且转化为list的模式
with open(filename, 'rt') as csvfile:
lines = csv.reader(csvfile)
dataSet = list(lines)
for x in range(len(dataSet) - 1):
for y in range(4):
dataSet[x][y] = float(dataSet[x][y])
#这里random随机生成小于1的数,spilt实际上代表了划分比例
if random.random() < split:
trainSet.append(dataSet[x])
else:
testSet.append(dataSet[x])
第二步,获取testData中点的k个临近点,其中涉及到一个计算点到点的距离,这里不仅仅限制在2,3维
# 多维计算距离///,length可以表示维度
def euclideanDistance(instance1, instance2, length):
distance = 0
for x in range(length):
print(type(instance1[x]), instance1[x])
print('----------------------')
print(instance2[x])
distance += pow((instance1[x] - instance2[x]), 2)
return math.sqrt(distance)
def getNeighbors(trainingSet, testInstance, k):
distances = []
# 获取未知点的维度
length = len(testInstance) - 1
for x in range(len(trainingSet)):
# 数据集里到目标点的距离
dist = euclideanDistance(testInstance, trainingSet[x], length)
distances.append((trainingSet[x], dist))
#表示根据第0个元素排序
distances.sort(key=operator.itemgetter(1))
neighbors = []
# 取前k个距离
for x in range(k):
neighbors.append(distances[x][0])
return neighbors
第三步,对k个临近点分类,找出对多的那一类
def getResponse(neighbors):
classVotes = {}
for x in range(len(neighbors)):
response = neighbors[x][-1]
if response in classVotes.keys():
classVotes[response] += 1
else:
classVotes[response] = 1
sortedVotes = sorted(classVotes.items(), key=lambda classVotes: classVotes[1], reverse=False)
return sortedVotes[-1]
第四步,验证,查看正确率
def getAccuracy(testSet, predictions):
print('--------------------------------------------')
print(predictions)
correct = 0
for x in range(len(testSet)):
print(testSet[x][-1], predictions[x])
if testSet[x][-1] == predictions[x]:
correct += 1
return (correct / float(len(testSet))) * 100.0
第五步:定义主函数:
def main():
trainSet = []
testSet = []
split = 0.67
loadDataset(r'iris.data.txt', split, trainSet, testSet)
predictions = []
k = 3
for x in range(len(testSet)):
neighbors = getNeighbors(trainSet, testSet[x], k)
result = getResponse(neighbors)
predictions.append(result[0])
print('>predicted=' + repr(result) + ',actual=' + repr(testSet[x][-1]))
accuracy = getAccuracy(testSet, predictions)
print("Accuracy: " + repr(accuracy) + '%')