CTR_stacking

The following codes are the solutions (5st place, private score: 0.78882) for the dacon competition.

1. Environmental settings

1.1 Clone this repository

git clone https://github.com/GNOEYHEAT/CTR_stacking.git
cd CTR_stacking

1.2 Install packages

conda create -n clicking python==3.9
conda activate clicking
pip install -r requirements.txt

1.3 Directory Structure

|-- README.md
|-- data
|   |-- data.zip
|   |-- pp_test_ce.parquet
|   |-- pp_train_ce.parquet
|   |-- sample_submission.csv
|   |-- test.csv
|   |-- train.csv
|-- requirements.txt
|-- src
|   |-- Preprocess.py
|   |-- config.py
|   |-- model.py
|   `-- run.py

2. Data Preprocessing

cd ./src
python Preprocess.py

3. Train Model

python run.py

4. Submission

The final submission is submission/stack.csv