Datasets Hub
Pointer page for shared datasets and teaching extracts used across the predictive site — Titanic, actuarial listings, house prices, and more.
A dataset is a promise about what was observed. Document the promise; then model. Skip the first step and your AUC is theater.
Hub contents
Dataset index
Teaching extracts for actuaries practicing predictive analytics: Titanic, listings, and the case-study files.
Teaching extracts
Small, clean slices for demos and exam-style case studies — prefer these over scraping production dumps into notebooks.
Provenance habit
Every extract should name source, as-of date, and row grain. Future-you will thank present-you.
How to use this hub
- Name the source. The teaching extracts live in QuantumData. Case studies on this site cite the same grain.
- Match grain to question. Claim-level vs policy-year vs household — wrong grain is the silent killer of “significant” results.
- Version lightly. If you clean a CSV, keep the raw beside the cleaned copy with a one-line changelog.
- Connect to case studies. Actuarial exams, house prices, vehicle safety, and the modeling app all drink from curated extracts nearby.
The dataset repository is QuantumData. This page is the on-site index.