|
|
from sklearn.datasets import load_iris
|
|
|
from sklearn.model_selection import train_test_split
|
|
|
from sklearn.naive_bayes import GaussianNB
|
|
|
from sklearn.tree import DecisionTreeClassifier
|
|
|
from sklearn.preprocessing import StandardScaler
|
|
|
from sklearn.neighbors import KNeighborsClassifier
|
|
|
import numpy as np
|
|
|
|
|
|
|
|
|
def bayes_iris(x):
|
|
|
"""
|
|
|
朴素贝叶斯对鸢尾花种类进行预测
|
|
|
:param x: 预测数据
|
|
|
:return:
|
|
|
"""
|
|
|
|
|
|
#1）获取数据集
|
|
|
iris = load_iris()
|
|
|
# print("查看特征值的名字：\n", iris.feature_names)
|
|
|
# print("查看特征值：\n", iris.data, iris.data.shape)
|
|
|
|
|
|
#2）划分数据集
|
|
|
x_train,x_test,y_train,y_test = train_test_split(iris.data,iris.target,random_state=22)
|
|
|
|
|
|
#3）建立模型
|
|
|
estimator= GaussianNB();
|
|
|
|
|
|
#4）模型训练
|
|
|
estimator.fit(x_train,y_train)
|
|
|
|
|
|
#5）模型评估
|
|
|
#对比真实值：
|
|
|
y_predict = estimator.predict(x_test)
|
|
|
# print("y_predict：\n",y_predict)
|
|
|
# print("直接对比真实值：\n",y_predict==y_test)
|
|
|
|
|
|
#计算准确率
|
|
|
score = estimator.score(x_test,y_test)
|
|
|
print("贝叶斯模型准确率为：\n",score)
|
|
|
|
|
|
#7）做出预测
|
|
|
#[0]表示setosa，[1]表示versicolor，[2]表示virginica
|
|
|
X_new = np.array(x)
|
|
|
prediction = estimator.predict(X_new)
|
|
|
print("贝叶斯模型预测的目标类别是：{}".format(prediction))
|
|
|
print("贝叶斯模型预测的目标类别花名是：{}".format(iris['target_names'][prediction]))
|
|
|
print()
|
|
|
|
|
|
return None
|
|
|
|
|
|
def decisiontree_iris(x):
|
|
|
"""
|
|
|
决策树对鸢尾花种类进行预测
|
|
|
:param x: 预测数据
|
|
|
:return:
|
|
|
"""
|
|
|
|
|
|
# 1) 获取数据集
|
|
|
iris = load_iris()
|
|
|
|
|
|
# 2）划分数据集
|
|
|
x_train, x_test, y_train, y_test = train_test_split(iris.data, iris.target, random_state=22,test_size=0.3)
|
|
|
|
|
|
# 3） 决策树预估器
|
|
|
estimator = DecisionTreeClassifier(criterion="entropy")
|
|
|
estimator.fit(x_train, y_train)
|
|
|
|
|
|
# 4） 模型评估
|
|
|
# 1、直接对比真实值和预测值
|
|
|
y_predict = estimator.predict(x_test)
|
|
|
# print("y_predict:\n", y_predict)
|
|
|
# print("比较:\n", y_test == y_predict)
|
|
|
|
|
|
# 2、计算准确率
|
|
|
score = estimator.score(x_test, y_test)
|
|
|
print("决策树模型准确度为:\n", score)
|
|
|
|
|
|
# 5) 做出预测
|
|
|
X_new = np.array(x)
|
|
|
prediction = estimator.predict(X_new)
|
|
|
print("决策树模型预测目标类别是:{}".format(prediction))
|
|
|
print("决策树模型预测目标花名是:", iris["target_names"][prediction])
|
|
|
print()
|
|
|
|
|
|
return None
|
|
|
|
|
|
def knn_iris(x):
|
|
|
"""
|
|
|
KNN近邻对鸢尾花种类进行预测
|
|
|
:param x:测试数据
|
|
|
:return:
|
|
|
"""
|
|
|
# 1) 获取数据
|
|
|
iris = load_iris()
|
|
|
|
|
|
# 2）划分数据集
|
|
|
x_train, x_test, y_train, y_test = train_test_split(iris.data, iris.target, random_state=22)
|
|
|
|
|
|
# 3）特征工程：标准化
|
|
|
transfer = StandardScaler()
|
|
|
x_train = transfer.fit_transform(x_train)
|
|
|
x_test = transfer.transform(x_test)
|
|
|
|
|
|
# 4) KNN算法预估器
|
|
|
estimator = KNeighborsClassifier(n_neighbors=3)
|
|
|
estimator.fit(x_train, y_train)
|
|
|
|
|
|
# 5）模型估计
|
|
|
# 方法1：直接比对真实值和预测值
|
|
|
y_predict = estimator.predict(x_test)
|
|
|
# print("y_predict:\n", y_predict)
|
|
|
# print("直接比对真实值和预测值:\n", y_test ==y_predict)
|
|
|
|
|
|
#方法2：计算准确值
|
|
|
score = estimator.score(x_test, y_test)
|
|
|
print("KNN模型准确值为：\n", score)
|
|
|
|
|
|
# 6) 做出预测
|
|
|
X_new = np.array(x)
|
|
|
prediction = estimator.predict(X_new)
|
|
|
print("KNN模型预测目标类别是:{}".format(prediction))
|
|
|
print("KNN模型预测目标花名是:", iris["target_names"][prediction])
|
|
|
print()
|
|
|
|
|
|
return None
|
|
|
|
|
|
def input_feature():
|
|
|
"""
|
|
|
顺序输入花萼长度、宽度，花瓣长度、宽度
|
|
|
:return:测试值二维数组
|
|
|
"""
|
|
|
feature = [0] * 4
|
|
|
name = ['Sepal Length：','Sepal Width：','Petal Length：','Petal Width：']
|
|
|
predict = []
|
|
|
count = 0
|
|
|
for i in range(4):
|
|
|
try:
|
|
|
feature[i] = float(input(name[i]))
|
|
|
except ValueError:
|
|
|
print("请输入数值")
|
|
|
return None
|
|
|
else:
|
|
|
count += 1
|
|
|
predict.append(feature)
|
|
|
return predict
|
|
|
|
|
|
if __name__ == '__main__':
|
|
|
X_new = input_feature()
|
|
|
if X_new != None:
|
|
|
decisiontree_iris(X_new)
|
|
|
knn_iris(X_new)
|
|
|
bayes_iris(X_new) |
