机器学习项目实战——集成预测政治献金

# 导入相关库
import numpy as np
import pandas as pd

import matplotlib.pyplot as plt
%matplotlib inline

import os
os.environ['PATH'] += ';...\\Graphviz2.38\\bin'  # Graphviz 临时环境变量

数据预处理

### 导入数据
# 设置随机种子
SEED = 222
np.random.seed(SEED)

df = pd.read_csv(r'...\data\federal_giving.csv', low_memory=False)
# 去除整列值相同的列
df = df.loc[:, (df != df.iloc[0]).any()]

feature_columns = ['entity_tp', 'classification', 'rpt_tp', 'cycle',
                   'transaction_amt', 'state', 'transaction_tp']

### 训练集和测试集
from sklearn.model_selection import train_test_split
from sklearn.metrics import roc_auc_score

def get_train_test(test_size=0.95):  # 速度快方便学习
    """划分训练集测试集"""
    y = 1 * (df['cand_pty_affiliation'] == "REP")  # 映射 REP:1, 非REP:0
    X = df[feature_columns].copy()

    # 填充空值 mode()众数
    for col in ['cycle', 'entity_tp', 'state']:
        X[col].fillna(X[col].mode()[0], inplace=True)

    X = pd.get_dummies(X, sparse=True)  # 独热编码，内存要求高

    return train_test_split(X, y, test_size=test_size, random_state=SEED)

xtrain, xtest, ytrain, ytest = get_train_test()

特征值说明

• entity_tp: 个人还是组织
• classification: 领域
• rpt_tp: 贡献的大小
• cycle: 捐赠年份
• transaction_amt: 捐赠金额
• state: 州

查看目标值的种类和数量

df['cand_pty_affiliation'].value_counts(normalize=True).plot(
   kind="bar", title="Share of No. donations")
plt.show()

定义绘制决策树的方法

import pydotplus
from IPython.display import Image
from sklearn.metrics import roc_auc_score
from sklearn.tree import DecisionTreeClassifier, export_graphviz

def print_graph(dtr, feature_names):
    """绘制决策树"""
    graph = export_graphviz(dtr, label="root", proportion=True, impurity=False, 
                            out_file=None, feature_names=feature_names, class_names={
   0: "D", 1: "R"},
                            filled=True, rounded=True)
    graph = pydotplus.graph_from_dot_data(graph)  
    return Image(graph.create_png())

绘制决策树

# 创建最大深度3的决策树对象
dtr1 = DecisionTreeClassifier(max_depth=3, random_state=SEED)
dtr1.fit(xtrain, ytrain)
# 预测概率值
p1 = dtr1.predict_proba(xtest)[:, 1]
# 决策树的 ROC-AUC 得分
print("Decision tree ROC-AUC score: %.3f" % roc_auc_score(ytest, p1))
# 绘制决策树
print_graph(dtr1, xtrain.columns