Analytics
by @ops Python + Postgres: Research Data Pipeline with Provenance Tracking
A data pipeline that records exactly where every dataset and transformation came from, a worker processing runs and logging lineage — distinct from generic ETL: built for research/pharma contexts where reproducibility and provenance are the actual requirement, not just moving data.
Loads this recipe straight into the wizard — every setting stays editable before you generate anything.
The stack
Install
# pulls this exact recipe straight from the Vault $ npx stacktora install python-research-data-provenance --remote $ stacktora sync # generate the files
What you get
More from the Vault
BI Reporting & Warehouse Query Layer
Fast aggregate queries over ClickHouse for internal dashboards — distinct from ETL (which moves data in), Reverse ETL (which pushes data out), and Scheduled Reports (which generates and delivers files): this is live query serving for a BI tool or internal dashboard.
Customer Data Platform / Unified Profile Backend
Merges customer data from multiple sources into one resolved profile via a worker doing the identity resolution — distinct from generic analytics: this is specifically about having one true record per customer, not aggregate reporting.
Customer Health Score & Churn Signals
Engagement events rolled up into a health/churn-risk score per account, computed by a worker on a schedule rather than live on every dashboard load. For B2B SaaS teams who need to know which accounts are quietly disengaging before the cancellation email arrives.
Deno + Postgres + Redis: URL Shortener & Link Analytics
Short-link generation, redirect handling, and click analytics — Redis for the redirect lookup that needs to be near-instant, Postgres for the click history behind it.