Skip to content

Instantly share code, notes, and snippets.

@zredlined

zredlined/generate_datasets.py

Last active February 19, 2021 06:41
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save zredlined/7db2dc22b3f9a29c290449f2e848aa37 to your computer and use it in GitHub Desktop.
Save zredlined/7db2dc22b3f9a29c290449f2e848aa37 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
generate_datasets.py
# Use the model to generate synthetic datasets
seed_data = json.loads(trends_df[seed_fields].to_json(orient="records"))
for dataset in range(5):
model.generate(max_invalid=1e5,
num_proc=1, # disable paralellism when using seed data
seed_fields=seed_data)
df = restore_daily(model.get_synthetic_df(), start, trend_col)
df.to_csv(f'synthetic-data-{dataset}.csv', index=False)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment