synthetics_creditcard_fraud_config.py

#!pip install gretel-synthetics --upgrade

from gretel_synthetics.batch import DataFrameBatch
from pathlib import Path

config_template = {
    "max_lines": 0,
    "max_line_len": 2048,
    "epochs": 7,
    "vocab_size": 20000,
    "gen_lines": 1000,
    "dp": False,
    "field_delimiter": ",",
    "overwrite": True,
    "checkpoint_dir": str(Path.cwd() / "checkpoints")
}

# train synthetic model
batcher = DataFrameBatch(df=training_set, batch_size=32, config=config_template)
batcher.create_training_data()
batcher.train_all_batches()

# generate synthetic dataset
status = batcher.generate_all_batch_lines(max_invalid=5000)
df_synthetic = batcher.batches_to_df()

# only keep fraudulent records created by our model
df_synthetic = df_synthetic[df_synthetic['Class'] == 1]