机器学习-数据预处理及特征工程

第八课 机器学习简单介绍
第十节 数据预处理及特征工程
# 引入包
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
准备数据集
# 加载数据
iris_data = pd.read_csv('./dataset/iris.csv')

# 获取特征
X = iris_data[['sepal_length', 'sepal_width', 'petal_length', 'petal_width']].values

# 获取标签
y = iris_data['label'].values

# 划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=1/3, random_state=10)
特征范围归一化
from sklearn.preprocessing import MinMaxScaler

scaler = MinMaxScaler()
# 在训练集上进行归一化
X_train_scaled = scaler.fit_transform(X_train)
# 在测试集上使用相同的scaler进行归一化
X_test_scaled = scaler.transform(X_test)
可视化归一化前后的结果
import matplotlib.pyplot as plt

fig, axs = plt.subplots(1, 2, figsize=(8, 8))

# 归一化前
axs[0].scatter(X_train[:, 0], X_train[:, 1])
axs[0].set_aspect('equal')
axs[0].set_xlabel('sepal length')
axs[0].set_ylabel('sepal width')
axs[0].set_title('Before scale')



axs[1].scatter(X_train_scaled[:, 0], X_train_scaled[:, 1])
axs[1].set_aspect('equal')
axs[1].set_xlabel('sepal length')
axs[1].set_ylabel('sepal width')
axs[1].set_title('After scale')
plt.show()

模型选择
from sklearn.neighbors import KNeighborsClassifier

# K近邻距离算法
knn_model = KNeighborsClassifier(n_neighbors=1)
训练模型
# 在未归一化的训练集上进行训练
knn_model.fit(X_train, y_train)

# 进行测试
knn_model.score(X_test, y_test)
0.96
# 在归一化的训练集上进行训练
knn_model.fit(X_train_scaled, y_train)

# 进行测试
knn_model.score(X_test_scaled, y_test)
0.98