Altair AI Hub -- for teams!
It's a mantra in computer science that you should use "the best tool for the job". This observation is often followed by the comment that the best tool for the job is the one you know well, because with it you can quickly and efficiently put your skills to use.
But what's the best tool for a team? Ideally it should be flexible enough to encompass a variety of tools, while unifying them onto a single platform to boost productivity. Maybe you can recognize yourself in one of the following categories:
- Do you like Python? Altair AI Hub includes JupyterLab and smoothly integrates with Python workflows.
- Already using Altair AI Studio? Altair AI Hub provides a shared platform, running on more powerful hardware, to enhance the team's productivity.
Altair AI Hub is a scalable, secure, easy-to-use platform. It provides a single shared workspace for team collaborations, with numerous productivity-enhancing features:
- Project management tools to keep track of the work that's been done
- Sharing of data / processes / models when appropriate, with fine-grained control of users and groups
- Schedules and queues for long-running processes on powerful server hardware
It's one thing to create a lot of models, quite another to deploy them to get useful predictions! The data workflows that your team creates will be more useful if the output can be consumed in a convenient fashion. Altair AI Hub provides Panopticon and Web API endpoints.
Continue reading to learn more, or investigate the items in the side menu.
- Create a data workflow / model
- Consume the output of a data workflow / model
- Web API endpoints
- Productivity tools
Create a data workflow / model
RapidMiner provides two different tools for creating a data workflow: Altair AI Studio is a standalone product, while JupyterLab is a service provided by Altair AI Hub.
- Altair AI Studio is a visual workflow designer for ETL and model building
- JupyterLab is a user-friendly data interface for users of R, Python and Anaconda
Altair AI Studio
Altair AI Studio is a visual workflow designer that enables more complex data flows, including ETL and model building. Altair AI Hub enhances Altair AI Studio by providing shared repositories for models and processes, batch jobs, scheduling, and project management.
JupyterLab
JupyterLab, a service included in Altair AI Hub, provides a user-friendly interface for data analysis using R, or Python and Anaconda. Note that Python users have full access to Projects and other productivity tools provided by Altair AI Hub, and can easily integrate their work.
Consume the output of a data workflow / model
To monitor and evaluate the output of your data workflows, Altair AI Hub includes Panopticon and Web API endpoints. A programmatic interface is available via a REST API and a Python API.
Panopticon
Altair AI Hub includes Panopticon, supporting a wide range of information visualizations, including Treemaps, Heat Maps, Scatter Plots, Horizon Graphs, and other visualizations designed for fast comprehension and easy interpretation of static, time series, real-time streaming, and historic data sets.
Web API endpoints
You can use Web API endpoints to remotely trigger RapidMiner processes, make predictions, and read the output in a browser or any other software that knows HTTP!
Productivity
Altair AI Hub improves productivity by providing a shared workspace for re-use of connections, data, workflows, code, models and results. In this shared workspace:
- Projects provide step-by-step work history
- Users and groups provide access control
- Schedules and queues provide smooth sharing of hardware resources
Projects
Altair AI Hub provides Projects so that the team will have a step-by-step history of the work they have done. The Projects back end is provided by Git, but no knowledge of Git is required to use this feature.
Users and Groups
Altair AI Hub provides role-based access control via users and groups. It can integrate with any SAML v2.0 and OAuth2 capable identity provider for identity federation, and with LDAP servers for user federation.
Schedules and queues
Long-running jobs can be dispatched more quickly and efficiently on server hardware. A queue system helps to organize these jobs, so that everyone gets their turn and so that the server never sleeps. If timing is important, your jobs can be scheduled, or triggered by an external event.
