Exam AWS Certified Machine Learning - Specialty topic 1 question 366 discussion

Copied from ChatGPT Given the constraints of minimizing operational overhead and the need for testing, the best deployment approach is to utilize the testing and evaluation features that are built into Amazon Forecast. Amazon Forecast allows for backtesting, which uses historical data to evaluate the performance of a forecasting model. This built in functionality provides the testing that is asked for, without the overhead of setting up a complex deployment strategy. Therefore, the correct combination is: A. Develop the model by using the Amazon Forecast Prophet model. B. Develop the model by using the Amazon Forecast holidays featurization and weather index. Because these two options are used together, and Amazon forecast has built in testing, there is no need to add another deployment option.

A. No, porque usar Prophet sin incorporar la featurización integrada de feriados y clima no optimiza la predicción para este caso. B. Sí, porque Amazon Forecast con featurización de feriados y clima integra esos factores de forma nativa y reduce el esfuerzo de modelado. C. Sí, porque una estrategia canary con SageMaker y Step Functions permite probar el modelo de forma gradual y con poco esfuerzo operativo. D. No, porque implementar A/B testing con SageMaker Pipelines añade complejidad adicional innecesaria para una prueba de pocos días. E. No, porque la estrategia A/B con SageMaker y Step Functions implica más sobrecarga operativa en comparación con una implementación canary.

Company want to test it for 2-3 days

BE for relevant features, E for simplicity

These steps ensure that the model is developed with relevant features and deployed in a way that minimizes operational risks and overhead.