TabTreeFormer: Tabular Data Generation Using Hybrid Tree-Transformer

Pre-requisites

• Python>=3.9 installed.
• pip install -r requirements.txt.

Usage

To train,

python main.py train -d DATA_PATH -t TARGET_COLUMN -p TASK_TYPE -o OUT_DIR

After training, to generate,

python main.py sample -c OUT_DIR -n N_ROWS -o OUT_CSV_PATH

For instance, to train and sample iris dataset (one can get the dataset by sklearn.datasets.load_iris), one can run the following:

python main.py train -d iris.csv -t target -p mult -o out
python main.py sample -c out -n 150 -o synthetic-iris.csv

b0d7cdb (first commit)