tencent_2017_contest_final http://algo.tpai.qq.com/home/information/index.html (2017腾讯社交广告算法大赛)

---

dependency

sudo pip install lightgbm pandas numpy sklearn fire

---

useage

./run_phase_1.sh

need: 8G RAM; 50G disk; 2 hour time;

初赛的脚本只需 8G 运行内存，　　 50G 硬盘空间，　 整个流程只需不到 2 个小时

or

./run.sh

need: 8G RAM; 500G disk; 10 hour time;

决赛的脚本同样只需 8G 运行内存, 500G 硬盘空间，　整个流程需要大约 10 个小时

---

notice

the run.sh or run_phase_1.sh will auto download the data, auto preprocess, auto extract feature, auto run model

run.sh 或 run_phase_1.sh 会自动下载数据，自动预处理，自动提取特征，自动跑模型生成submiss.csv.

---

---

other solutions

final 14th https://github.com/z564808896/Tencent_Social_Ads

final 23th https://github.com/BladeCoda/Tencent2017_Final_Coda_Allegro