lianyhaii 0.0.14

A package to win data competition

lianyhaii

usage

第一步：安装与测试

pip install lianyhaii
import lianyhaii
print(lianyhaii.__version__)

第二步：准备数据、特征、参数

import pandas as pd 
import numpy  as np 
import lianyhaii
import warnings 
import sys 
warnings.filterwarnings('ignore')

## 定义数据集、label、训练参数、特征名
train = pd.DataFrame({
    'y':(np.random.random(size=500)>0.5).astype(int)
})
test = pd.DataFrame({
    'y':(np.random.random(size=500)>0.5).astype(int)
})
for i in range(10):
    train[f'x{i}'] = np.random.random(size=500)
    test[f'x{i}'] = np.random.random(size=500)
label = 'y'
lgb_params = {
    'objective': 'binary',
    'boosting_type': 'gbdt',
    'metric': 'auc',
    'early_stopping_rounds': 50,
    'verbose':-1,
}
base_features = [f'x{i}' for i in range(10)]

以上是送进模型的主要准备工作，接下来会轻松不少

第三步：快速模型测试

mt = make_test(train,test,base_features=base_features,new_features=[],
            m_score=[[0.0,]],label=label,metrices=['auc'],log_tool=None)
mt.init_CV(seed=412,CV_type='skFold',n_split=5)
oof,pred = mt.lgb_test(lgb_params=lgb_params)
## 得到oof和pred方便后续调整或者提交

获奖经历

2021 山东赛 公积金贷款逾期预测 A榜 rank2/xxx B榜 rank6/xxx
2021 梧桐杯 5G潜客识别 B榜 rank7/xxx
2022 招商银行Fintech rank23/xxx -> 开源