MLS-C01 Question 45
Select 4You are a machine learning engineer tasked with building a real-time fraud detection system for a financial services company. The system needs to process high-throughput streaming transaction data, apply a machine learning model for fraud detection, and store flagged transactions in Amazon DynamoDB for further analysis. You decide to use Amazon Managed Service for Apache Flink. Which features of Apache Flink should you leverage to meet these requirements?
- A
Use Flink SQL to query and process the incoming streaming data.
- B
Leverage Apache Flink's built-in machine learning libraries to train the fraud detection model directly within the Flink application.
- C
Integrate a pre-trained machine learning model with Apache Flink using custom operators for real-time inference.
- D
Use Amazon Kinesis Data Streams as a source to ingest real-time transaction data into Apache Flink.
- E
Set up a sink connector in Apache Flink to write flagged transactions directly to Amazon DynamoDB.
Show answer and explanation
Correct answers: A, C, D, E
Explanation
Amazon Managed Service for Apache Flink enables processing high-throughput streaming data using features like Flink SQL for querying, integration with pre-trained ML models for real-time inference, and native support for source and sink connectors such as Amazon Kinesis Data Streams and Amazon DynamoDB. While Apache Flink is powerful for stream processing, training machine learning models is typically done outside the Flink environment and integrated after training. This makes options 1, 3, 4, and 5 correct for this scenario.
- A. Correct.
Flink SQL is an essential feature of Apache Flink that allows you to process streaming data using SQL-like queries. This is highly relevant for filtering and transforming the transaction data.
- B. Incorrect.
Apache Flink's built-in machine learning libraries are generally not used for training models within the application. Instead, pre-trained models are typically integrated for real-time inference.
- C. Correct.
You can integrate a pre-trained machine learning model into Apache Flink using custom operators or APIs for real-time inference. This approach is efficient for applying the fraud detection model to streaming data.
- D. Correct.
Amazon Kinesis Data Streams is a common and scalable source for ingesting high-throughput streaming data into Amazon Managed Service for Apache Flink.
- E. Correct.
Apache Flink supports sink connectors, including writing data to Amazon DynamoDB. This is appropriate for storing flagged transactions for later analysis.