We are happy to share the new capabilities that improve the BigML platform: operating thresholds and organizations.
Operating thresholds, available from the BigML Dashboard, API, and WhizzML, allow Machine Learning practitioners to fine tune the performance of any classification model. By setting the right operating threshold, you can be more or less aggressive when predicting a class for a given instance of your dataset. This process is especially relevant in domains such as fraud detection, loan risk prediction, and medical diagnosis, where the consequences of some classifications may have prohibitive costs associated.
With the increased regulatory focus on avoiding negative consumer outcomes due to digital transformation and decision automation initiatives, it becomes even more critical to understand the tradeoffs between different operating modes of predictive models, and the smart applications in which they operate. Ignorance or following a black box approach are not viable options when a single bad decision can cost millions of dollars. Operating thresholds help companies avoid such scenarios by controlling the tradeoff between false positives and false negatives to minimize risks and costs.
Organizations, the second feature released, presents a convenient collaborative space that makes it easy and more efficient for companies to adopt Machine Learning across their entire enterprise. Organizations help professionals from several departments to work on the same Machine Learning projects, from different accounts at different permission levels.
If you missed the live webinar or you want to watch it again, please visit the BigML YouTube channel.
For further learning on operating thresholds and organizations, please visit our dedicated release page, where you will find:
- The series of six blog posts that gradually explain operating thresholds.
- An extra blog post about how organizations can improve the productivity of several teams in your company.
- The detailed documentation to learn how to use operating thresholds from the BigML Dashboard, the API, and WhizzML.
- The slides used during the webinar.
As usual, remember that you can always reach out to us at firstname.lastname@example.org for any feedback or questions. Thanks again for all your positive comments on the webinar!