You work as a machine learning specialist for a hedge fund firm. Your traders trade in highly volatile securities and derivatives. In real-time, their trading activity must be monitored for an anomalous activity to keep the firm from entering into potentially very large high-risk transactions that could jeopardize the firm’s valuation and collateral obligations with the Securities and Exchange Commission (SEC). Which of the following options best describes your optimal machine learning solution to this problem?

Answer options:

A.Use Kinesis Data Streams to gather the trading, valuation, and collateral data from your investment management systems, source the data from Kinesis Data Streams toKinesis Data Analytics, use SQL to transform the data and write it to S3, use the SageMaker Random Cut Forest built-in algorithm to detect anomalous trading activity.
B.Use Kinesis Data Firehose to gather the trading, valuation, and collateral data from your investment management systems, source the data from Kinesis Data Firehose toKinesis Data Analytics, use SQL to transform the data and write it to S3, use the SageMaker k-means built-in algorithm to detect anomalous trading activity.
C.Use Kinesis Data Streams to gather the trading, valuation, and collateral data from your investment management systems, source the data from Kinesis Data Streams toKinesis Data Analytics, use Apache Flink to transform the data and write it to S3, use the SageMaker Random Cut Forest built-in algorithm to detect anomalous trading activity.
D.Use a Glue ETL job to gather the trading, valuation, and collateral data from your investment management systems. Have the Glue ETL job transform the data and write it to S3, use the SageMaker Random Cut Forest built-in algorithm to detect anomalous trading activity.

Correct Answer: C
Option A is incorrect. Kinesis Data Streams will satisfy your real-time data gathering requirement. The SageMaker Random Cut Forest algorithm will satisfy your anomaly detection requirement. However, using SQL in your Kinesis Data Analytics application will require a Lambda function to write your data to S3.
Option B is incorrect. Kinesis Data Firehose will deliver your data in near real-time, not real-time like Kinesis Data Streams. Also, using SQL in your Kinesis Data Analytics application will require a Lambda function to write your data to S3. Finally, the k-means algorithm is not the best choice for anomaly detection. Random Cut Forest is the best algorithm for anomaly detection.
Option C is correct. Kinesis Data Streams will satisfy your real-time data gathering requirement. Kinesis Data Analytics running an Apache Flink application will allow you to transform your data and directly write it to S3. Finally, the Random Cut Forest algorithm is the best choice for anomaly detection.
Option D is incorrect. A Glue ETL job will not satisfy your real-time data delivery requirement. Glue ETL is used in batch applications.
References:
Please see the US Securities and Exchange Commission compliance alert dated July, 2008 (https://www.sec.gov/about/offices/ocie/complialert0708.htm),
The Amazon Kinesis Data Analytics developer guide titled Example: Writing to an Amazon S3 Bucket (https://docs.aws.amazon.com/kinesisanalytics/latest/java/examples-s3.html),
The Amazon SageMaker developer guide titled Random Cut Forest (RCF) Algorithm (https://docs.aws.amazon.com/sagemaker/latest/dg/randomcutforest.html)