XGBoost baseline

XGBoost是一个优化的分布式梯度增强库,它在Gradient Boosting框架下实现机器学习算法。XGBoost提供了并行树提升（也称为GBDT，GBM）。
使用交叉验证，以f1为评价方法的baseline:

#!/usr/bin/env python
# _*_ coding:utf-8 _*_

import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import xgboost as xgb
from sklearn.externals import joblib
from sklearn.model_selection import StratifiedKFold
from sklearn.metrics import f1_score
import warnings
warnings.filterwarnings('ignore')

pd.set_option('display.max_columns', 500, 'display.width', 1000)  # 设置显示宽度
data_train = pd.read_csv("train_3.csv")
data_test = pd.read_csv("test_3.csv")
# 对current_service映射编码
label2current_service = dict(zip(range(0, len(set(data_train['current_service']))),
                                 sorted(list(set(data_train['current_service'])))))
current_service2label = dict(zip(sorted(list(set(data_train['current_service']))),
                                 range(0, len(set(data_train['current_service'])))))
data_train['current_service'] = data_train['current_service'].map(current_service2label)
# print(len(set(data_train['current_service'])))      # 15种类型
y = data_train.pop('current_service')
user_id = data_train.pop('user_id')
x_train = data_train
test_user_id = data_test.pop('user_id')
x_test = data_test
print(x_train.info())
X, y, X_test = x_train.values, y.values, x_test.values
n_splits = 5
seed = 2333


# 采取k折模型方案
# 自定义F1评价函数
def f1_score_vail(pred, data_vail):
    labels = data_vail.get_label()
    score_vail = f1_score(y_true=labels, y_pred=pred, average='macro')      # xgb的predict输出即为对应的label
    return '1-f1_score', 1-score_vail   # xgb目标是将目标指标降低


# xgb参数
xgb_params = {
    "max_depth": 7,
    "learning_rate": 0.05,
    "objective": "multi:softmax",
    "silent": 1,
    "eta": 0.1,
    "n_jobs": -1,
    "num_class": 15,
    "subsample": 0.8,
    "min_child_weight": 2,
    "seed": 2333,
    "alpha": 0.1,
    "lambda": 0.2,
    "predictor": "gpu_predictor",
    "num_boost_round": 100000,
    "colsample_bytree": 0.7,
    "colsample_bylevel": 0.7,
    "tree_method": 'gpu_exact'
}

x_score = []
cv_pred = []
skf = StratifiedKFold(n_splits=n_splits, random_state=seed, shuffle=True)
X_test = xgb.DMatrix(X_test)    # 转化为xgb需要的数据格式

for index, (train_index, test_index) in enumerate(skf.split(X, y)):
    print(index)
    X_train, X_valid, y_train, y_valid = X[train_index], X[test_index], y[train_index], y[test_index]
    train_data = xgb.DMatrix(X_train, y_train)      # 训练集
    validation_data = xgb.DMatrix(X_valid, y_valid)     # 验证集
    watchlist = [(validation_data, 'train')]    # 验证的数据
    clf = xgb.train(xgb_params, train_data, num_boost_round=100000, early_stopping_rounds=100, feval=f1_score_vail,
                    evals=watchlist, verbose_eval=1)    # 训练
    joblib.dump(clf, "model/xgb_{}.m".format(index))
    # clf = joblib.load("model/xgb_{}.m".format(index))
    X_valid = xgb.DMatrix(X_valid)  # 转为xgb需要的格式
    x_pred = clf.predict(X_valid)
    x_score.append(f1_score(y_valid, x_pred, average='macro'))
    y_test = clf.predict(X_test)
    if index == 0:
        cv_pred = np.array(y_test).reshape(-1, 1)
    else:
        cv_pred = np.hstack((cv_pred, np.array(y_test).reshape(-1, 1)))
    if index == 4:
        xgb.plot_importance(clf, max_num_features=25)
        plt.title("Feature_importance")
        plt.show()
cv_pred = cv_pred.astype(np.int64)  # 转为int64,不然后面会报错
# 投票
submit = []
for line in cv_pred:
    submit.append(np.argmax(np.bincount(line)))
# 保存结果
df_test = pd.DataFrame()
df_test['id'] = list(test_user_id.unique())
df_test['predict'] = submit
df_test['predict'] = df_test['predict'].map(label2current_service)
df_test.to_csv('output/xgb.csv', index=False)
print(x_score, np.mean(x_score))