AI-900 Question 80
Select 3You are tasked with building a machine learning solution on Azure to predict customer churn for a subscription-based service. You need to identify the fundamental principles of machine learning to guide your approach. Which of the following are essential considerations when designing the solution?
- A
Ensure the data is clean, relevant, and representative of the problem you are solving.
- B
Select the most complex machine learning algorithm available to maximize accuracy.
- C
Split the data into training, validation, and testing datasets to evaluate model performance.
- D
Use pre-built Azure ML models without any modifications to save time.
- E
Monitor the model for performance degradation over time and retrain as needed.
Show answer and explanation
Correct answers: A, C, E
Explanation
The fundamental principles of machine learning on Azure include ensuring data quality, properly splitting datasets for training and evaluation, and implementing ongoing monitoring and maintenance of models. These practices help ensure the success of machine learning solutions. Simply selecting complex algorithms or using pre-built models without consideration for the problem specifics do not align with fundamental machine learning principles.
- A. Correct.
Data quality is a fundamental prerequisite for any machine learning solution. Clean, relevant, and representative data ensures the model can learn meaningful patterns.
- B. Incorrect.
While algorithm selection is important, choosing the most complex algorithm is not always the best approach. Simpler models may perform well and are easier to interpret.
- C. Correct.
Splitting data ensures that you can properly train the model while evaluating its performance on unseen data, which is a crucial step in the machine learning lifecycle.
- D. Incorrect.
While pre-built models can be useful for quick deployments, they may not always align with the specific requirements of your problem and may require customization.
- E. Correct.
Monitoring model performance ensures that it remains accurate over time. Retraining is necessary when the model begins to underperform due to changes in data patterns (concept drift).