Analytics that ends in a decision

Seven projects. One reviewable analytical workflow.

Classification, forecasting, experimentation, SQL, segmentation, financial planning, and healthcare operations—now delivered as a fast, credential-free Vercel experience.

Python-derived snapshotsInteractive browser toolsSource limitations visible
Analytics projects07Distinct business workflows
Static routes08Direct links and refreshes
Runtime API keys0No browser secrets
Hosting runtimeStaticNo Streamlit server required
Featured work

Choose a business problem to explore

Each project connects the analytical method to a decision, exposes its assumptions, and preserves useful interaction in the browser.

01Interactive case study

Customer Churn & Retention

Prioritize retention outreach with a reproducible classifier, adjustable threshold, and customer-level risk scenario.

ClassificationROC–AUC 0.886Thresholds
Open project →
02Interactive case study

U.S. Retail Sales Forecast

Backtest a seasonal forecast against a year-ago baseline and examine uncertainty over the planning horizon.

Time series36.4% MAE gainBacktesting
Open project →
03Interactive case study

A/B Test Decision Calculator

Evaluate conversion lift, confidence intervals, practical significance, expected impact, and required sample size.

ExperimentationConfidence intervalsPower
Open project →
04Interactive case study

E-commerce SQL Analytics

Analyze marketplace orders with safe SQL grains, cohort retention, delivery KPIs, and category performance.

DuckDB SQLCohortsData quality
Open project →
05Interactive case study

Customer Segmentation

Build transparent RFM audiences, validate a clustering challenger, and turn segments into measurable campaign tests.

RFMK-meansPower BI
Open project →
06Interactive case study

Financial Planning & Variance

Reconcile a real operating budget, investigate Q3 pacing, and test revenue, cost, and hiring scenarios.

FP&AExcelBudget vs actual
Open project →
07Interactive case study

Healthcare Access & Readiness

Model state access and staffing in DuckDB, then screen review priorities with visible data completeness.

HealthcareDuckDBPublic data
Open project →
Migration architecture

The analytics stays reviewable. The delivery gets simpler.

01

Python analysis

Original notebooks, models, SQL, and validation code remain in the repository for reproducibility.

02

Reviewed snapshots

An explicit exporter transforms model results and reviewed aggregates into deterministic JSON.

03

Interactive Next.js

Browser tools handle thresholds, filters, scenarios, statistical tests, and safe CSV downloads.

04

Static Vercel hosting

No long-running Streamlit process, runtime database, persistent storage, or client-side secret is required.