Python数据分析与机器学习在金融风控中的应用小结

import pandas as pd
import sqlite3
# 连接到SQLite数据库
conn = sqlite3.connect('financial_records.db')
# 查询客户交易记录
query = '''
SELECT transaction_id, customer_id, transaction_amount, transaction_date, transaction_type
FROM transactions
'''
df = pd.read_sql_query(query, conn)
# 关闭数据库连接
conn.close()
# 查看数据
print(df.head())

# 数据清洗：去除重复记录
df = df.drop_duplicates()
# 处理缺失值：填充或删除缺失值
df = df.fillna(method='ffill')
# 标准化：将数值型特征标准化到相同的尺度
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
df[['transaction_amount']] = scaler.fit_transform(df[['transaction_amount']])
# 查看预处理后的数据
print(df.head())

# 提取特征和标签
X = df[['customer_id', 'transaction_amount', 'transaction_type']]
y = df['default']
# 将类别特征进行独热编码
X = pd.get_dummies(X, columns=['transaction_type'])
# 查看提取后的特征
print(X.head())

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)
# 查看划分后的数据集
print(X_train.shape, X_test.shape)

from sklearn.ensemble import RandomForestClassifier
# 初始化随机森林模型
model = RandomForestClassifier(n_estimators=100, random_state=42)
# 训练模型
model.fit(X_train, y_train)
# 查看模型训练效果
print(model)

from sklearn.metrics import accuracy_score, recall_score, f1_score
# 预测测试集
y_pred = model.predict(X_test)
# 计算评估指标
accuracy = accuracy_score(y_test, y_pred)
recall = recall_score(y_test, y_pred)
f1 = f1_score(y_test, y_pred)
# 输出评估结果
print(f'准确率：{accuracy}')
print(f'召回率：{recall}')
print(f'F1分数：{f1}')

# 假设已经有市场行情数据的DataFrame
market_data = pd.read_csv('market_data.csv')
# 处理缺失值
market_data = market_data.fillna(method='ffill')
# 标准化
scaler = StandardScaler()
market_data[['price']] = scaler.fit_transform(market_data[['price']])
# 查看预处理后的数据
print(market_data.head())

# 提取特征和标签
X = market_data[['price', 'volume']]
y = market_data['price'].shift(-1)  # 预测下一个时间点的价格
# 去除空值
X = X[:-1]
y = y.dropna()
# 查看提取后的特征
print(X.head())

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)
# 查看划分后的数据集
print(X_train.shape, X_test.shape)

from sklearn.linear_model import LinearRegression
# 初始化线性回归模型
model = LinearRegression()
# 训练模型
model.fit(X_train, y_train)
# 查看模型训练效果
print(model)

from sklearn.metrics import mean_squared_error, mean_absolute_error
# 预测测试集
y_pred = model.predict(X_test)
# 计算评估指标
mse = mean_squared_error(y_test, y_pred)
mae = mean_absolute_error(y_test, y_pred)
# 输出评估结果
print(f'均方误差：{mse}')
print(f'平均绝对误差：{mae}')

# 假设已经有操作风险数据的DataFrame
operation_data = pd.read_csv('operation_data.csv')
# 处理缺失值
operation_data = operation_data.fillna(method='ffill')
# 标准化
scaler = StandardScaler()
operation_data[['amount']] = scaler.fit_transform(operation_data[['amount']])
# 查看预处理后的数据
print(operation_data.head())

# 提取特征和标签
X = operation_data[['amount', 'operation_type', 'time']]
y = operation_data['risk']
# 将类别特征进行独热编码
X = pd.get_dummies(X, columns=['operation_type', 'time'])
# 查看提取后的特征
print(X.head())

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)
# 查看划分后的数据集
print(X_train.shape, X_test.shape)

from sklearn.tree import DecisionTreeClassifier
# 初始化决策树模型
model = DecisionTreeClassifier(random_state=42)
# 训练模型
model.fit(X_train, y_train)
# 查看模型训练效果
print(model)

# 预测测试集
y_pred = model.predict(X_test)
# 计算评估指标
accuracy = accuracy_score(y_test, y_pred)
recall = recall_score(y_test, y_pred)
f1 = f1_score(y_test, y_pred)
# 输出评估结果
print(f'准确率：{accuracy}')
print(f'召回率：{recall}')
print(f'F1分数：{f1}')