Data Science in Microsoft Fabric

Data Science and Machine Learning in Microsoft Fabric: Hands-on Training for Data Scientists and ML Engineers

Building a model is only part of the job. Getting it into production, keeping it healthy and governing it properly is where many data science projects stall. This hands-on course shows you how to take a machine learning solution through its whole lifecycle in Microsoft Fabric, from designing the data architecture to retraining a deployed model.

You will learn how the Fabric data science components fit together, how to prepare lakehouse and Delta tables for machine learning, and how to track experiments and register models with MLflow. You will then operationalise those models with batch scoring, monitoring and retraining, and promote them through environments using Git integration and deployment pipelines. The course finishes with AutoML and responsible AI, so you know when automation helps and how to govern it.

Develop Fabric Data Science Skills:

• How the Fabric data science architecture fits together, including its components and an end-to-end workflow.

• How to prepare data for ML using lakehouse tables, Delta tables, partitioning, optimisation and feature storage.

• How to choose between batch and streaming data for machine learning.

• How to choose between Spark and Pandas for model development.

• How to use MLflow for experiment tracking and logging, and how to register, version and describe models with metadata.

• How to operationalise models using the model registry, batch scoring, monitoring and retraining.

• How to use Git integration and deployment pipelines to manage and promote ML solutions between environments.

• How to use AutoML, when it is the right choice, and how to apply governance and responsible AI practices.

Learn Fabric Data Science Your Way, Training Tailored to Your Needs

This course is offered in a UK-based face-to-face classroom or virtual setting and can be tailored to your needs. If your team has an existing Fabric environment, a particular use case or specific governance requirements, let us know and we can focus the course on what matters most to you.

Solve Your Machine Learning Struggles with Expert Training

Whether you are a data scientist, ML engineer or data engineer supporting data science teams, those who come to PTR for this training are usually asking:

• How do I structure my data and architecture so models are easy to build and maintain?

• How do I keep track of experiments and know which model version is in production?

• How do I get a model out of a notebook and into a repeatable scoring process?

• How do I know when a model has degraded, and how do I retrain it?

• How do I move ML work safely between development and production?

• When should I use AutoML, and how do I keep it responsible and governed?

If this sounds familiar, this is the course for you. We will walk you through practical answers to each of these, with exercises you can apply straight away in your own environment.

This course is aimed at people who already build or support machine learning solutions. Delegates should have:

• Working knowledge of Python

  , including using notebooks and common data libraries such as Pandas.

• A basic understanding of machine learning concepts

  , such as training and testing data, model evaluation and common algorithms. You do not need to be an expert.

• Some familiarity with Microsoft Fabric

  , including lakehouses and notebooks, is helpful but not essential.

• A basic understanding of Git

  is helpful for the enterprise ML session.

Prior experience of Spark, Delta tables and MLflow is not required, as these are introduced during the course.

To take part in the hands-on exercises, delegates will need access to a Microsoft Fabric environment where they can create lakehouses, notebooks and ML items. We can provide this or help you set up a trial. If you are unsure whether your experience is the right fit, get in touch and we will be happy to advise.