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

Model not able to predict anomaly after more than few data points #129

Open
msplmansit98 opened this issue May 29, 2023 · 2 comments
Open

Comments

@msplmansit98
Copy link

I was exploring Luminaire, But after using the opt_config from hyperopt and creating the model, the model is not able to predict the anomaly status after a certain data point
It shows-
{'Success': False, 'ErrorMessage': 'Current trained model object expired'}
I have a daily data of 4.5k-5k data points
What am I missing here ?
Why for some data points anomaly status is not showing up ?
What is the min and max data range that can be feed + how much it can predict ?

@sayanchk
Copy link
Collaborator

sayanchk commented Jun 1, 2023

@msplmansit98 Luminaire is a time series anomaly detection tool that tracks temporal patterns and variations over time. The reliability of the model depend on the fact that it needs to be trained at a regular cadence. Therefore, by design, Luminaire has a model expiry concept (ModelFreshness score) that expires a model if an outdated model object is being used for serving. Please refer to the outlier detection documentation for more details.

@sayanchk
Copy link
Collaborator

sayanchk commented Jun 1, 2023

@msplmansit98 If you have daily data with 4.5k-5k data points, that means your observations are coming at a higher frequency than an hour. Luminaire batch model currently supports till hourly cadence (reference). You might want to look into the streaming data module which is more flexible and could be a better option for your use case.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants