νμμ΄ νΌ λ¬Έμ 리μ€νΈμ μ λ΅ μ¬λΆκ° λ΄κΈ΄ λ°μ΄ν°λ₯Ό λ°νμΌλ‘ νμμ μ§μμνλ₯Ό μΆμ νκ³ , λ―Έλμ νμμ΄ νΉμ λ¬Έμ λ₯Ό λ§μΆμ§ ν릴μ§λ₯Ό μμΈ‘ν©λλ€. μ΄λ₯Ό ν΅ν΄ νμμκ² κ°μΈ λ§μΆ€ν κ΅μ‘μ μ 곡ν©λλ€.
AUROC, Accuracy
>>> cd code
>>> python train.py --wandb_project_name [PROJECT_NAME] --wandb_run_name [RUN_NAME] --model [MODEL]
>>> cd code
>>> python inference.py --wandb_run_name [RUN_NAME] --model [MODEL]
>>> cd code
>>> python train_kfold.py --wandb_project_name [PROJECT_NAME] --wandb_run_name [RUN_NAME] --model [MODEL] --kfold 10
>>> cd code
>>> python inference_kfold.py --wandb_run_name [RUN_NAME] --model [MODEL] --kfold 10
>>> cd code
>>> python train_stfkfold.py --wandb_project_name [PROJECT_NAME] --wandb_run_name [RUN_NAME] --model [MODEL] --kfold 10
>>> cd code
>>> python inference_kfold.py --wandb_run_name [RUN_NAME] --model [MODEL] --kfold 10
- LSTM (lstm)
- LSTM + Attention (lstmattn)
- Bert (bert)
- GRUATTN (gruattn)
- ATTNGRU (attngru)
- Saint (saint)
- Saint_custom (saintcustom)
- LastQuery (lastquery)
- BaseCNN (cnn)
- DeepCNN (deepcnn)
βββ README.md - 리λλ―Έ νμΌ
β
βββ requirements.md - νμν library
|
βββ dkt/ - DLν utils νμΌ
β βββ criterion.py
β βββ custom_model.py
β βββ dataloader.py
β βββ feature.py
β βββ model.py
β βββ optimizer.py
β βββ scheduler.py
β βββ trainer.py
| βββ utils.py
|
βββ code/ - DLν μ½λ ν΄λ
β βββ args.py
β βββ inference.py
β βββ inference_kfold.py
β βββ train.py
β βββ train_kfold.py
| βββ train_stfkfold.py
β
βββ notebook_pycaret - MLν μ½λ ν΄λ
| βββ Add_Feature_with_Groupby.ipynb
| βββ get_logsμ°μ΅ν΄λ³΄κΈ°.ipynb
| βββ kaggle_riid_μ μ²λ¦¬.ipynb
| βββ LGBM_Validμνλλλ‘ꡬμΆμ±κ³΅.ipynb
β βββ Optuna_LightGBM_λ¬Έμ μκ°κ°κ²©νμ²λ¦¬X.ipynb
β βββ outputνμΌ_bestλλΉκ΅ν΄λ³΄κΈ°.ipynb
| βββ PermutationImportance.ipynb
|
βββ notebooks
βββ baseline.ipynb
βββ EDA_Minyong.ipynb
βββ EDA-arabae.ipynb
βββ hard_and_soft_ensemble.ipynb
βββ output_confidence.ipynb
βββ Riiid_pretrain.ipynb
κ°λ―Όμ© T1001 [Github] [Blog]
- EDA
- GRU + Attention & SAINT Modeling
- User Data Augmentation, Pseudo Labeling
- Deep Learning Code κ°μ
- DKT, DKT+, DKVMN λ Όλ¬Έ μ 리 λ° κ³΅μ
κΉμ§ν T1248 [Github] [Blog]
- EDA
- Saint, Saint+ Modeling
- Feature Searching (λ¬Ένλ³ λμ΄λ / KnowledgeTag)
- Ensemble (Hard + Soft voting)
λ¬Έμ¬ν T1058 [Github] [Blog]
- ML (with Customized Optuna & Pycaret)
- κ²μ¦μ λ΅ (HoldOut Set, Customized CV)
- Efficient Feature Engineering (with Pandas method)
- Feature Selection (with Permutation Importance)
- Riiid Dataset Processing for Pre-training
λ°°μλΌ T1084 [Github] [Blog]
- LSTM, LSTM+Attention, BERT, CNN, Last Query, SAINT λ± λ€μν λͺ¨λΈ ꡬν λ° μ€ν
- Userλ³ Feature Engineering
- Deep Learning Code κ°μ
- Riiid λ°μ΄ν°λ₯Ό νμ©ν pre-train μλ
- Ensemble (soft-voting, weighted soft-voting)
μ΄μ ν T1160 [Github] [Blog]
- EDA
- RNNκ³μ΄(LSTM, LSTM+Attention) λͺ¨λΈ μ€ν
- DKT, DKVMN λ Όλ¬Έμ 리
- Riiid Competition Data Analysis for Transfer Learning
- νμ λ΄μ© μ 리
μ΅μ λΌ T1212 [Github] [Blog]
- EDA - νμ΅ λ°μ΄ν°, ν μ€νΈ λ°μ΄ν° λΆν¬ νμ
- Feature engineering - νμ΄ μκ°, μ νλ νκ· feature μΆκ°
- ML λͺ¨λΈ νμ΅ - LightGBM, XGBoost, CatBoost
- Validation set μ°ΎκΈ°