Churn Prediction — Leakage-Free Retention Model
Leakage-free churn-модель: recall@top-10% = 0.53 и lift 3,07× при ROC-AUC 0.904 на хронологическом сплите — без утечки будущей активности в признаки.
PythonLightGBMscikit-learnpandas / NumPymatplotlibpytestuv
Эффект: Chronological snapshot split (train/val/test) — no future activity leaks into training, Churn label = a future 30-day inactivity window for recently active users only, Decision metric recall@top-10% = 0.53, lift@top-10% = 3.07x, LightGBM Brier 0.068 vs 0.099 for the balanced logistic baseline
Cohort Analysis Dashboard
Треугольная когортная матрица удержания и LTV на синтетических данных: ARPU и LTV по когортам с поправкой на observation age, выгрузка в Tableau (CSV + Hyper).
Pythonpandasmatplotlib / seabornJupyter NotebookTableau (Hyper API)
Эффект: Cohort retention matrix with triangular decay, ARPU / LTV by cohort with proper observation-age caveat, Tableau-ready export (CSV + .hyper extract), Reproducible seeded pipeline (seed=42)
Browser Mini-Games — Analytics Arcade
10 играбельных мини-игр в self-contained SVG: 7 аналитических (A/B до p<0.05, funnel drop, cohort catch, retention day) и 3 аркадных. Один файл — вся игра, без сборки.
SVGJavaScriptAstroPlaywright
Эффект: 10 игр: 7 аналитических + 3 аркадных, каждая — self-contained SVG, Аналитические концепции → игровая механика: p<0.05, retention day, funnel bottleneck, Zero-dependency: HTML+CSS+JS в одном файле, без сборки и сервера, Phone (swipe/tap) + desktop (keyboard/mouse), 3 темы, PostHog-трекинг (game_selected) + CTA → контакт
TaskFlow — PostHog Product Analytics Pipeline
SaaS-продукт, инструментированный PostHog end-to-end: типизированный каталог событий, генерация трафика и 7 анализов — воронка, retention, A/B, revenue/LTV, time-to-convert.
PythonPostHogFastAPIStreamlitpandas / matplotlibpytest / ruffDocker
Эффект: Typed event catalog (single source of truth) + PostHog capture/identify/group with PII scrubbing, Feature flag → onboarding A/B variant; A/B analysis with chi-square, uplift, Wilson CI + SRM check, Day-N cohort retention, time-to-convert, revenue/LTV, first-feature → upgrade conversion, The same metrics as SQL (BI / interview reference) + interactive Streamlit dashboard, CI (pytest + ruff) + Docker + render.yaml for one-click deploy
SQL Analytics Case Study
25 SQL-кейсов на синтетическом датасете (~183k событий): воронка, retention, LTV, атрибуция, аномалии. Запуск на DuckDB одной командой, живой отчёт на GitHub Pages.
SQLDuckDBPythonpandas / NumPypytest
Эффект: 25 self-contained SQL cases (funnel → RFM), DuckDB — no server, no credentials, one command, Regression tests with deterministic invariants per case, Synthetic deterministic data (seed=42 + additive seed=43), Live interactive report on GitHub Pages
Product Analytics Dashboard (Streamlit)
Продуктовый дашборд на синтетическом SaaS-датасете (8 000 пользователей): AARRR-воронка, cohort retention, выручка (MRR, ARPU, churn) и сегментация — 5 страниц на Streamlit.
PythonStreamlitpandas / NumPy
Эффект: 8,000 synthetic users, Jan 2024 – Jun 2025, deterministic seed = 42, AARRR funnel: app_open → signup → activate → start_trial → subscribe with step drop-off, Cohort retention heatmap (signup month × months since signup), Revenue: MRR growth, MRR by plan, ARPU, logo churn, Segments: distributions + conversion + ARPU by segment / channel / country / device
