AWS · MLA-C02 · BETA
Data sources, formats, and storage
MLA-C02 · Lesson 1 of 23 · 4:39
YouTube loads only after you press play. Watch on YouTube ↗
In this lesson
Lesson 1 of the free MLA-C02 beta series (AWS Certified Machine Learning Engineer – Associate, domain 1, data preparation). In this lesson: where training data comes from, choosing storage by cost, performance, structure, and compliance (Amazon S3, EFS, RDS, DynamoDB), row and column formats like CSV, JSON, Parquet, and ORC, and capacity, throttling, and request rates. With three quick checks and a recap, and AnyCompany Mobility, a car sharing service, as the running example.
Preview the topics
- Three years of trip records sit in S3. Data scientists read only the last six months, weekly; older records stay for audits, read about once a year. What is the most cost-effective storage?
- Data scientists run Athena queries on 400 columns of car telemetry, stored as JSON files in S3. Each query uses only five columns, and query costs keep growing. What should the ML engineer do?
- Car telemetry arrives in a Kinesis stream in provisioned mode. At evening peaks, producers get throughput exceeded errors, although the stream is far below its capacity. The partition key is the city, and one city sends half the traffic. What should the ML engineer do first?