PROFESSIONAL-MACHINE-LEARNING-ENGINEER Question #208: Real Exam Question with Answer & Explanation

Question

You work as an analyst at a large banking firm. You are developing a robust scalable ML pipeline to tram several regression and classification models. Your primary focus for the pipeline is model interpretability. You want to productionize the pipeline as quickly as possible. What should you do?

Options

• AUse Tabular Workflow for Wide & Deep through Vertex AI Pipelines to jointly train wide linear
• BUse Google Kubernetes Engine to build a custom training pipeline for XGBoost-based models
• CUse Tabular Workflow for TabNet through Vertex AI Pipelines to train attention-based models
• DUse Cloud Composer to build the training pipelines for custom deep learning-based models

Explanation

To quickly productionize a robust, scalable ML pipeline for regression and classification with a primary focus on model interpretability, use Tabular Workflow for TabNet through Vertex AI Pipelines.

Common mistakes.

• A. While Wide & Deep models offer some interpretability due to their linear component, TabNet is specifically highlighted for its inherent interpretability through attention, making it a stronger choice when interpretability is the primary focus.
• B. Building a custom training pipeline on Google Kubernetes Engine (GKE) requires substantial configuration and management effort, which contradicts the goal of productionizing the pipeline 'as quickly as possible'.
• D. Using Cloud Composer for custom deep learning models involves significant development and integration work for the models themselves, which might not be the quickest path to production when interpretability is a primary focus.

Concept tested. Vertex AI Tabular Workflows for interpretable models

Reference. https://cloud.google.com/vertex-ai/docs/pipelines/tabular-workflows/tabnet

No community discussion yet for this question.