import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report, confusion_matrix
import xgboost as xgb

# 模拟一个简单的交易数据集
data = {
    'transaction_amount': [100, 200, 150, 50, 300, 400, 120, 80],
    'transaction_time': ['2025-02-05 12:00:00', '2025-02-05 13:30:00', '2025-02-05 15:00:00',
                         '2025-02-05 16:30:00', '2025-02-06 12:00:00', '2025-02-06 13:00:00',
                         '2025-02-06 14:30:00', '2025-02-06 16:00:00'],
    'user_id': ['user1', 'user2', 'user3', 'user1', 'user2', 'user3', 'user1', 'user2'],
    'device_info': ['device123', 'device124', 'device123', 'device125', 'device126', 'device124', 'device123', 'device125'],
    'ip_address': ['IP123', 'IP124', 'IP125', 'IP126', 'IP127', 'IP124', 'IP123', 'IP126'],
    'is_fraud': [0, 1, 0, 0, 1, 0, 0, 1]  # 1 表示欺诈，0 表示正常
}

# 创建 DataFrame
df = pd.DataFrame(data)

# 提取特征矩阵 X 和标签 y
X = df[['transaction_amount', 'user_id', 'device_info', 'ip_address']]  # 选择特征列
y = df['is_fraud']  # 标签是是否欺诈

# 对类别特征进行编码（如 user_id, device_info, ip_address 等）
X = pd.get_dummies(X, columns=['user_id', 'device_info', 'ip_address'])

# 将 transaction_time 转换为数值（如将时间转化为时间戳）
X['transaction_time'] = pd.to_datetime(df['transaction_time']).view('int64') / 10**9  # 转换为 Unix 时间戳

# 拆分数据集为训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# 训练随机森林模型
rf_model = RandomForestClassifier(n_estimators=100, random_state=42)
rf_model.fit(X_train, y_train)

# 预测测试集
y_pred_rf = rf_model.predict(X_test)

# 评估模型
print("Random Forest Classification Report:")
print(classification_report(y_test, y_pred_rf))
print(confusion_matrix(y_test, y_pred_rf))

# 使用 XGBoost 训练
dtrain = xgb.DMatrix(X_train, label=y_train)
dtest = xgb.DMatrix(X_test, label=y_test)

params = {
    'objective': 'binary:logistic',
    'eval_metric': 'logloss',
    'max_depth': 6,
    'learning_rate': 0.1,
    'n_estimators': 100
}

# 训练模型
xgb_model = xgb.train(params, dtrain, num_boost_round=100)

# 预测
y_pred_xgb = xgb_model.predict(dtest)
y_pred_binary_xgb = [1 if p > 0.5 else 0 for p in y_pred_xgb]

# 评估模型
print("XGBoost Classification Report:")
print(classification_report(y_test, y_pred_binary_xgb))
print(confusion_matrix(y_test, y_pred_binary_xgb))

# 新的交易数据
new_transaction = [[100, 'user1', 'device123', 'IP123', 1644052800]]  # 使用 Unix 时间戳
new_transaction = pd.get_dummies(pd.DataFrame(new_transaction, columns=['transaction_amount', 'user_id', 'device_info', 'ip_address', 'transaction_time']))

# 预测新交易是否为欺诈
predicted_label_rf = rf_model.predict(new_transaction)
predicted_label_xgb = xgb_model.predict(xgb.DMatrix(new_transaction))

# 输出预测结果
if predicted_label_rf == 1:
    print("随机森林模型预测：交易存在欺诈风险！")
else:
    print("随机森林模型预测：交易正常。")

if predicted_label_xgb > 0.5:
    print("XGBoost模型预测：交易存在欺诈风险！")
else:
    print("XGBoost模型预测：交易正常。")