TruSHAP is an easy introduction to TruEra designed with SHAP users in mind. Using the TruSHAP extension, you can add your model to a TruEra deployment with just 2 code changes to your notebooks that already use SHAP.
This SHAP Extension allows you to get familiar with TruEra ingestion and Trustworthy ML concepts while using the model-agnostic SHAP interface. Creating the explainer in TruSHAP automatically ingests your model and data to your TruEra application without any additional lines of code.
Once you get up and running with TruSHAP, transitioning to the TruEra way of analyzing models comes easy. All you have to do is add your TruEra credentials, the TruSHAP API does the rest.
If you don’t have TruEra credentials yet, get them instantly for free.
Sign up at: https://app.truera.net/
All you need to do to unlock TruEra capabilities is to:
- Change the import from import shap to: import truera.client.experimental.trushap
- Add your connection_string and token as arguments to shap.Explainer()
Once you’ve made those two changes, the model will be added to a new project and data collection in your TruEra deployment.
When you use the explainer, e.g. explainer(X), the data split X will also be added to your TruEra deployment.
And all your existing SHAP code still works, TruSHAP only augments the SHAP Explainer.
- TruSHAP is Faster Than Making Tea: Watch the Video