Analytics foundation
Statistical reasoning supports experimentation, KPI interpretation, model evaluation, variance analysis, and confidence in product and business decisions.
What this system does
Interprets experiments, variance, distributions, correlation, confidence, and sampling limits.
Supports A/B testing, funnel changes, churn analysis, KPI movement, model evaluation, and operational review.
Makes analysis more credible by asking what changed, what uncertainty exists, and what action is justified.
How I use it
Product analytics examples include conversion, churn, NPS, and CSAT metrics that require interpretation, not just calculation.
Python churn workflow reports RMSE and ROC AUC to evaluate baseline model behavior.
BudgetDB QA checks use variance thresholds and pass/review statuses before reporting.
Related projects
Evidence or examples
Portfolio SQL demo calculates NPS, CSAT, churn rate, and conversion rate.
BudgetDB QA query compares source totals against modeled fact totals and labels pass/review status.
Churn script uses RMSE and ROC AUC to evaluate predictive output.
Define hypothesis -> choose metric -> estimate sample needs -> run test -> inspect lift, uncertainty, and guardrail metrics.CASE WHEN ABS(source_total - fact_total) <= 1 THEN 'PASS' ELSE 'REVIEW' END AS qa_statusCorrelation can identify relationships to investigate; it does not prove causation without stronger design or validation.