2018年甜橙金融杯大数据建模大赛初赛方案：通过追踪时间、设备、ip和经纬度等属性的变化来建模判断UID是否为黑产链
## 代码说明：
- gen_stat_feat.py　　　统计特征
- gen_w2v_feat.py　　　 word2vec特征
- lgb_train.py　lgb训练模型

两份特征建模加权8:2比例融合即可0.792+，单独统计特征加UID列建模即可0.795。(The preliminary scheme of the 2018 Sweet Orange Financial Cup Big Data Modeling Competition: Modeling and judging whether UID is a black chain by tracking the changes of attributes such as time, equipment, ip, longitude and latitude　　　　　　
## Code descr iption:　　　　　
- Statistical characteristics of gen_stat_feat.py　　　　　
- Characteristics of gen_w2v_feat.py word_2vec　　　　　　
- lgb_train.py LGB training model　　　　　　

The weighted 8:2 scale fusion of two feature models can achieve 0.792+, and the single statistical feature plus UID column can achieve 0.795.)