CTR經典模型如:FM,FFM,Wide&Deep,建議自己去復現一個完整的通用模型
先從pytorch版本入手(后期考慮tensorflow),從kaggle上找實際的比賽
github推薦:
https://github.com/mJackie/RecSys CTR相關資料匯總
https://github.com/duboya/CTR-Prediction/tree/46c303986ec57092d9eb4478a3583fa019d18efd/Algorithm%20Practice/GBDT%20%2B%20LR
GBDT+LR , XGBoost等
github 相關代碼推薦:
1.(第二推薦)
https://github.com/JianzhouZhan/Awesome-RecSystem-Models pytorch,tf2.0實現的常見ctr模型
Criteo's Kaggle display advertising challenge
2.
https://github.com/Johnson0722/CTR_Prediction tf實現常見ctr模型
比賽:https://www.kaggle.com/c/avazu-ctr-prediction/data
avazu
3. (第一推薦)
https://github.com/chenxijun1029/DeepFM_with_PyTorch pytorch,deepfm
比賽: https://www.kaggle.com/c/criteo-display-ad-challenge/data
Criteo's Kaggle display advertising challenge
解析源碼的博客:
https://zhuanlan.zhihu.com/p/84526966
4.
https://github.com/shenweichen/DeepCTR tf1.X,tf2.X,經典推薦系統代碼實現全集
5.
https://github.com/GitHub-HongweiZhang/prediction-flow pytorch,經典推薦系統代碼實現集合
