Drag and Drop Question You need to define an evaluation strategy for the crowd sentiment models. Which three actions should you perform in sequence? To answer, move the appropriate actions from the…

The correct answer is Add new features for retraining supervised models.; Evaluate the changes in correlation between model error rate and centroid distance; Filter labeled cases for retraining using the shortest distance from centroids. This question tests understanding of iterative evaluation strategies for hybrid clustering-supervised sentiment models, specifically how centroid distance metrics guide feature engineering and data selection for retraining.

Train and deploy models

Question

Drag and Drop Question You need to define an evaluation strategy for the crowd sentiment models. Which three actions should you perform in sequence? To answer, move the appropriate actions from the list of actions to the answer area and arrange them in the correct order. Answer:

Exhibit

Answer Area

Drag items

Add new features for retraining supervised models.Filter labeled cases for retraining using the shortest distance from centroidsEvaluate the changes in correlation between model error rate and centroid distanceImpute unavailable features with centroid aligned modelsFilter labeled cases for retraining using the longest distance from centroidsRemove features before retraining supervised models.

Correct arrangement

Add new features for retraining supervised models.
Evaluate the changes in correlation between model error rate and centroid distance
Filter labeled cases for retraining using the shortest distance from centroids

Explanation

This question tests understanding of iterative evaluation strategies for hybrid clustering-supervised sentiment models, specifically how centroid distance metrics guide feature engineering and data selection for retraining.

Approach. Step 1 - adding new features - enriches the supervised model before evaluating impact. Step 2 - evaluating the correlation between model error rate and centroid distance - is the diagnostic core: if samples far from cluster centroids consistently have higher error rates, centroid distance is a reliable proxy for model uncertainty. Step 3 - filtering labeled cases using the shortest centroid distance - selects the most representative, archetypal samples from each cluster for retraining, because these core cases best capture the true signal of each sentiment class and produce the most stable model updates.

Concept tested. Centroid-distance-guided evaluation in semi-supervised/hybrid sentiment models: using the relationship between clustering geometry (centroid distance) and supervised model error as a feedback signal, then leveraging cluster centroids to curate high-quality retraining data rather than relying on outliers or arbitrary sampling.

Reference. Microsoft Azure ML / DP-100 - Crowd Sentiment Model Evaluation with Clustering Feedback Loops

Topics

#model evaluation#evaluation strategy#sentiment analysis#machine learning lifecycle

Community Discussion

No community discussion yet for this question.

Full DP-100 Practice