80%准确率的 lightgbm 框架试用 kaggle桑德兰银行目标客户预测

Santander Customer Transaction Prediction

kaggle的一个比赛，试题于  https://www.kaggle.com/c/santander-customer-transaction-prediction

 

我的初步解答：

import pandas as pd
import numpy as np
from sklearn.ensemble import GradientBoostingClassifier
from sklearn import  metrics
import lightgbm as lgb
from sklearn.metrics import roc_auc_score
from sklearn.model_selection import train_test_split
 
import time

train = pd.read_csv('/home/kesci/input/BK_TRAIN2242/train.csv')
train.head()
test = pd.read_csv('/home/kesci/input/BK_TEST3071/test.csv')
test.head()
y = train['target']
X=train

X.drop(['ID_code','target'],axis = 1, inplace=True)
X

import copy
test2=copy.deepcopy(test)
test2.drop(['ID_code'],axis = 1, inplace=True)
test2.head()

#######################LGM成品框架


start = time.clock()
 
print("Loading Data ... ")
 
train_x, valid_x, train_y, valid_y = train_test_split(X,y, test_size=0.2, random_state=1)   # 分训练集和验证集
 
X_train = train_x
y_train = train_y
X_test = valid_x
y_test = valid_y
 
 
# create dataset for lightgbm
lgb_train = lgb.Dataset(X_train, y_train)
lgb_eval = lgb.Dataset(X_test, y_test, reference=lgb_train)
# specify your configurations as a dict
params = {
    'boosting_type': 'gbdt',
    'objective': 'binary',
    'metric': {'binary_logloss', 'auc'},
    'num_leaves': 5,
    'max_depth': 6,
    'min_data_in_leaf': 450,
    'learning_rate': 0.1,
    'feature_fraction': 0.9,
    'bagging_fraction': 0.95,
    'bagging_freq': 5,
    'lambda_l1': 1,  
    'lambda_l2': 0.001,  # 越小l2正则程度越高
    'min_gain_to_split': 0.2,
    'verbose': 5,
    'is_unbalance': True
}
 
# train
print('Start training...')
gbm = lgb.train(params,
                lgb_train,
                num_boost_round=10000,
                valid_sets=lgb_eval,
                early_stopping_rounds=500)
 
print('Start predicting...')
 
preds = gbm.predict(test2, num_iteration=gbm.best_iteration)  # 输出的是概率结果
 
# 导出结果
threshold = 0.5
for pred in preds:
    result = 1 if pred > threshold else 0
 
        
end = time.clock()
print (("完成 用时："))
print ((end - start))    



predictions=[]
for pred in preds:
    result = 1 if pred > threshold else 0
    predictions.append(result)
    


df=copy.deepcopy(pd.DataFrame(test['ID_code']))


df['target']=predictions
df.to_csv('out2.csv')

print ('生成结果csv完成')

模型还有很多参数可以继续调节，若要进10%至少要 score到88%以上