Frequently Asked Questions

• What is the trust radar and what kind of information dos it summarize?

  The trust radar is althiqa’s main functionality. It aims at summarizing information contained in a wide range of metrics in a geometrical way. When morre than one model is passed to the trust radar, althiqa then agregate all the metrics ranks of each models to produce a final ranking. If more than one metric is selected, the user can define importance weights for each metric and the ranking will be adjusted accordingly.

• Where can I see how predefined metrics are computed?

  You can look at the code for each predefined metrics in our metric repo file.

• What if I want to change the way predefined metrics are computed ?

  You are very welcome to do so. You can copy the code of the metric you want to change from our metric repo file. You can paste the code on your notebook, edit your changes and once you are done, give a new name to the resulting custom metric and push it with the API so it can be displayed and be ready to use in the interface.

• How can I help?

  We are actively releasing new functionalities that will complement the trust radar. If you find that a functionality you use to report or communicate ML information to your stakeholders is missing in the althiqa service, please open an issue so that we can help on it. The API will evolve based on users’ needs.