Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Scalability to run large number of time-series without crashing #152

Open
ekurniawan-ispt opened this issue Apr 5, 2022 · 0 comments
Open
Assignees
Labels
enhancement New feature or request

Comments

@ekurniawan-ispt
Copy link

Description

I get frustrated that it crashes or runs for days when running greater number time series such as 2500 time series.
I am referring to from_pandas_dynamic function (which is the Dynamic NOTEARS implementation in Causalnex)

Context

I love this package and I use it for small number of time series without issues.
However, recently I have been involved in a project with large number of time series, eg around 2500 time series to discover the causal relationships.

Possible Implementation

I suggest for the team look into the possibilities of implementing PySpark MLLib for Spark matrices and Spark dataframes.

Possible Alternatives

I have started implementing Granger Causal test using Pyspark Dataframe, but I still want to see Bayesian Network solution by Causalnex package.

GabrielAzevedoFerreiraQB pushed a commit to GabrielAzevedoFerreiraQB/causalnex that referenced this issue Jun 7, 2022
Co-authored-by: philip_pilgerstorfer <[email protected]>
@oentaryorj oentaryorj added the enhancement New feature or request label Aug 25, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

3 participants