Skip to content

References: Synthetic Data Generation and AI-Assisted LRS Testing

  1. Synthetic Data - Wikipedia - Comprehensive overview of synthetic data techniques, distributional fidelity, and the privacy advantages that make synthetic xAPI cohorts a better choice than scrubbed real data for most testing.

  2. Large language model - Wikipedia - The technology behind Claude Code; explains how LLMs convert prompts into structured JSON output and the constraints that govern statement-generation quality.

  3. Load testing - Wikipedia - Covers ramp-up profiles, sustained throughput tests, and breakpoint testing — the patterns this chapter applies to LRS endpoints with synthetic statement streams.

  4. The Art of Application Performance Testing (2nd Edition) - Ian Molyneaux - O'Reilly Media - Practitioner-focused treatment of load-test design, ramp profiles, and result interpretation that grounds the synthetic-cohort strategy this chapter teaches.

  5. Statistical Rethinking (2nd Edition) - Richard McElreath - CRC Press - The statistical foundation for evaluating whether a synthetic learner cohort is "representative" — distribution comparison, posterior predictive checks, and the discipline that prevents synthetic data from quietly lying.

  6. Anthropic API Documentation - Anthropic - The reference for the Claude API used throughout this chapter, including prompt-caching, structured tool use for JSON output, and rate-limit considerations for batch generation runs.

  7. Claude Code Documentation - Anthropic - The CLI tool this chapter uses to author synthetic-statement scripts; covers the agent-loop model, tool permissions, and the workflows the synthetic-data examples are built on.

  8. Faker Library - joke2k - The Python library for generating realistic names, emails, and timestamps; commonly used as a deterministic component alongside an LLM that handles the verb/object choices.

  9. Locust Load Testing Tool - Locust Project - Python-based distributed load tester that pairs naturally with synthetic-statement generators for stress-testing an LRS at production scale.

  10. k6 Load Testing - Grafana Labs - JavaScript-scriptable load tester whose statement-generation scripts can share JSON fixtures with the textbook's xAPI client. Excellent for end-to-end load tests of the full pipeline.