Train MachineLearning Models with MLflow in Microsoft Fabric

Explore how to train machine learning models using frameworks in Microsoft Fabric, including scikit-learn for traditional tasks and PyTorch or TensorFlow for deep learning, plus Synapse ML for scalable pipelines.

Learn a traditional machine learning workflow: load data as a data frame, explore with visualizations, engineer features, split into train and test sets, train, and evaluate performance metrics.

Train a regression model with scikit-learn by splitting data into X train/test and Y train/test, fit a linear regression model, generate predictions, and evaluate metrics with MLflow tracking.

Learn how to use notebooks in Microsoft Fabric to train models with Spark Compute, enabling PySpark, Python, ML frameworks like scikit-learn, PyTorch, TensorFlow, and Pandas data frames.

Learn how MLflow, an open source library, tracks and manages machine learning experiments by logging parameters, metrics, and artifacts, enabling a unified workflow in Microsoft Fabric.

Explore how MLflow unifies the machine learning lifecycle, tackling experiment management, reproducibility, deployment consistency, model management, and library agnosticism to streamline production.

Discover how data scientists, MLOps professionals, data science managers, and prompt engineers use MLflow to manage experiments, deployment, and model lifecycles.

Explore the Microsoft Fabric interface, switch between Fabric and Power BI experiences, manage workspaces and one lake containing all your datasets, and monitor real-time data streams.

Create a workspace in fabric using the new workspace button or the workspace menu, name it mlflow workspace, and configure basics, then add data pipelines, notebooks, and reports.

Create and name a notebook in a Microsoft Fabric workspace, switch between code and markdown cells, run code, and download in ipynb, html, python, or text formats.

Load the diabetes dataset from Azure Open Data into a spark data frame, convert to a pandas data frame for MLflow workflows in Microsoft Fabric, enabling data exploration and prep.

Train a regression model to predict diabetes from features like age, sex, BMI, bp, and s1–s6, using train-test split, scikit-learn, and MLflow for tracking.

Explore MLflow in Microsoft Fabric to view and list experiments, retrieve by name, inspect runs, and compare models using metrics like the R2 score in the Fabric GUI.

Explore experiment metadata across runs in MLflow on Fabric, compare mean absolute error, mean squared error, and R2 score, view model files and run details, and tailor visualizations.

Understand the ml model file and its artifacts, including the model folder, conda yaml, model.pkl, and python env yaml, and see how MLflow auto logging enables predictions.

Save the best performing model from an experiment run as a new ML model and use it to generate predictions, applying its version and referencing the linked experiment details.

Save and name your notebook, download it for sharing, and manage active mlflow sessions to free resources in microsoft fabric, using scikit-learn to train and track models.

Celebrate completing the train machine learning models with MLflow in Microsoft Fabric course, showcasing skills in managing ML workflows, experiment tracking, and enterprise-grade model management, with your completion certificate.