AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Digital note-taking applications for recording meeting minutes during ML project planning

  Why it's wrong here

  Digital note-taking applications such as OneNote or Microsoft Word serve to capture unstructured meeting minutes, decisions, and action items during project planning. While documentation is an important part of the ML lifecycle, these tools are not executable code environments and lack the kernels, compute backends, and dataset connections that define Azure ML notebooks. Notebooks are interactive, code-centric workspaces for hands-on data exploration, not archival note repositories.

• ✓

  Interactive Jupyter notebook environments for data exploration and model prototyping by data scientists

  Why this is correct

  Azure Machine Learning notebooks are cloud-hosted Jupyter environments that run on compute instances, giving data scientists interactive access to Python or R kernels. These notebooks support inline code execution, data visualizations via matplotlib, and iterative experimentation, enabling rapid prototyping of models directly against Azure datasets. Unlike static documentation or passive viewers, they allow users to modify code, rerun cells, and evolve analyses in real time within the ML workspace.

• ✗

  Automated logging notebooks that record all model training metrics without code

  Why it's wrong here

  Azure ML does support automated metric tracking, but this is done through experiment tracking features like MLflow or the Run.log APIs, which require explicit user code within training scripts. Notebooks themselves are not automatic loggers; they execute code and can log metrics only when the data scientist writes the appropriate tracking calls. The notion of a notebook that records training metrics without any code conflates the interactive coding environment with Azure ML's backend experiment telemetry, which operates at a separate pipeline layer.

• ✗

  Read-only document viewers for reviewing completed ML experiment results

  Why it's wrong here

  Reviewing completed ML experiment results typically uses Azure ML Studio's experiment dashboards, metric charts, and model explanation widgets, which present static outputs like loss curves or confusion matrices. In contrast, a Jupyter notebook is a read-write, executable canvas where code cells can be altered and recomputed against live compute resources. Notebooks are not read-only viewers; they let data scientists modify preprocessing, retrain a model, or generate new visualizations on the fly, making them unsuitable for passive document review.

Go deeper

Related to this question