Mnubo Data Science Studio

Collaborative ML Tool for
Model Productization

Our Data Science studio is a collaborative tool that helps you develop, productize and share machine learning models on IoT data.

Mnubo Data Science Studio allows you to train, test, deploy and run models more efficiently.

Overview

Mnubo Data Science Studio: Built for Efficiency

Mnubo’s fully hosted notebooks and function-as-a-service architecture remove the complexity and cost of building and maintaining ML models. With Mnubo Data Science Studio, you don’t need to reformat your code to take it to an external tool. It only takes a few clicks to productize and share your models.

  • Develop

    Mnubo Data Science Studio provides a hosted and fully managed Python notebook development environment in which you can leverage your preferred ML libraries. Custom dependency sets can also be added to streamline the development of proprietary intellectual properties.

  • Train, Test & Version

    Train, test, and version models on live streaming data, and/or archived data. Mnubo Data Science Studio has a flexible sandbox and production environment which allows users to easily experiment and validate models.

  • Productize and Deploy

    Mnubo Data Science Studio packages models into a function-as-a-service container. Productize your models on-demand to internal or external applications through a one-click deployment process or store them in a blob store for easy retrieval and scheduling.

  • Schedule and Run

    Schedule the training and prediction of your ML models, and save prediction indexes for quick retrieval. Export your insights and predictions to third-party systems and solutions using a secure JSON REST API call.

  • Visualize and Monitor Performance

    Easily examine and benchmark KPIs for model performance. Develop custom KPIs or leverage native analytics standards to ensure your ML models and workflows continue to perform over time.

Features & Benefits

Mnubo Data Core automatically cleans and data structures based on ingestion rules, making it immediately available in the Mnubo Data Science Studio. This eliminates the need to build and maintain your own ETL process.

  • Securely query any archived or live streaming data
  • Simplify access to Mnubo IoT Analytics Suite with an extensive JSON query language

Mnubo Data Science Studio provides IoT-specific libraries to accelerate the development of models, while also providing access to popular external libraries.

  • Integrates with Mnubo IoT Analytics Suite
  • Easy access to external libraries through dependency sets

Mnubo Data Science Studio provides version control, scheduling of notebooks, blob storage, and ‘bring your own model’ capabilities to enhance teamwork and sharing of IP.

  • Centralize all IoT-specific IP in one repository and reuse/distribute when required.
  • Bring models that were built with other tools.
  • Store, retrieve, schedule and run your models when required.

Mnubo Data Science Studio provides a sandbox environment to test models against archived and/or live data. Push to production with a one-click deployment feature, eliminating complex deployment codes and DevOps pipelines. All deployments are tracked and available for historic audits through the native logging system.

  • Access a sandbox environment and test before you deploy.
  • Train, test and refine against archived and/or live streaming data.
  • Productize your models and deployment process in one-click with function-as-a-service.
  • Monitored and unified logging system.

Mnubo Data Science Studio provides controls to manage compute power utilization and save model results for rapid retrieval by external applications.

  • Assign the right number of CPUs and memory when running models and notebooks to balance speed versus cost.
  • Save prediction indexes for quick retrieval.

Resources