Content explorer

Explorer results

Nothing matched. Clear a filter or change the query.

• Projects Churn Prediction — Leakage-Free Retention Model

  A leakage-free churn model: recall@top-10% of 0.544 and 5.44x lift at 0.916 ROC-AUC, on a chronological split with no future activity leaking into features.

  Analytics · python · 2026 · 14 connections

  Connections: Churn Prediction — Leakage-Free Retention Model

  Related

  • Churn

  Shares topic

  • retention

  Similar

  • A/B Testing Methodology Toolkit
  • Airflow DAG Platform
  • Causal / Uplift — CUPED and Individual Treatment Effects
  • LinkedIn Content Analytics
  • ML Fundamentals — scikit-learn Pipeline
  • Python Analytics Playground
  • RAG Evaluation Harness — retrieval and generation metrics
  • SERP Rank Lab
  • SQL Analytics Case Study
  • Volta Neobank — Product Analytics

  Mentions term

  • Retention
• Projects Digital Garden — EQ in IT

  Notes linked into a network, not a feed: a digital garden about emotional intelligence in IT on Quartz v4 — backlinks and a graph turn note accumulation into a navigable structure.

  Engineering · 2026 · 3 connections
• Projects RFM Analysis of Bank Clients

  Segmented a bank's clients into 6 RFM segments: the top 20% hold 38.2% of revenue (Gini 0.324). Synthetic dataset — 1,983 clients, 9,492 transactions.

  Analytics · sql · 2026 · 7 connections
• Projects Product Analytics + A/B on Supabase

  Full-stack analytics on Supabase: an A/B experiment lifted conversion +11.1pp (p ≈ 1.7e-10, chi-square), with a Streamlit dashboard reading live data through Row Level Security.

  Product · sql · 2026 · 6 connections
• Projects LinkedIn Content Analytics

  Analytics on my own LinkedIn posts: one parser, four renderers. Formats are compared with confidence intervals — and noise is called noise, not a ranking.

  Analytics · python · 2026 · 11 connections
• Projects SERP Rank Lab

  Metasearch SERP analytics lab: inverse-propensity weighting corrects position bias, NDCG and revenue per search disagree, and always-valid mSPRT holds the false-positive rate under repeated looks.

  Analytics · python · 2026 · 12 connections
• Projects RAG Evaluation Harness — retrieval and generation metrics

  RAG pipeline with a real offline harness: recall@5 and MRR = 1.00 on 20 questions, correctness 0.37. TF-IDF retrieval, no API keys; faithfulness ≡1.00 — an extractive invariant, not a measurement.

  Analytics · python · 2026 · 20 connections
• Glossary A/B test

  An A/B test compares two versions of a product across randomly split groups, separating the effect of a change from the natural variance of a metric.

  Experiments · ab-testing · 2026 · 13 connections
• Glossary CAC and unit economics

  CAC is the cost of acquiring one customer. Together with LTV it sets the payback of unit economics and the ceiling on affordable traffic.

  Business · segmentation · 2026 · 6 connections
• Glossary Confidence interval

  A confidence interval is the range that would contain the true effect at a stated rate across repetitions — unlike a p-value, it shows both the effect size and the uncertainty.

  Statistics · ab-testing · 2026 · 6 connections
• Glossary Correlation and causation

  Correlation shows that two quantities move together; causation means one drives the other. A relationship with no mechanism or control does not prove an effect.

  Statistics · ab-testing · 2026 · 6 connections
• Glossary Guardrail metric

  A guardrail metric is one an experiment must not worsen: it catches collateral damage while the primary metric moves up.

  Experiments · ab-testing · 2026 · 5 connections
• Glossary Sample size

  Sample size is the number of observations needed to detect an effect of a given size at a chosen power and significance. It is computed before a test starts.

  Statistics · ab-testing · 2026 · 6 connections
• Glossary A/A test

  An A/A test serves the same experience to both groups and validates the tooling — split, metrics, and standard error — before the main experiment.

  Experiments · ab-testing · 2026 · 8 connections
• Glossary ARPU

  ARPU is the average revenue per user over a period. A basic input to LTV and unit economics, sensitive to the mix of the active base.

  Business · segmentation · 2026 · 12 connections
• Glossary Bootstrap

  The bootstrap estimates a statistic's distribution by resampling with replacement, without relying on asymptotic formulas or normality.

  Statistics · ab-testing · 2026 · 10 connections
• Glossary Churn

  Churn is the share of users who stop being active over a period. It is the flip side of retention and a direct input to LTV.

  Product · retention · 2026 · 17 connections
• Glossary Cohort

  A cohort is a group of users sharing a common start time or event. It is the foundation of cohort-based retention analysis.

  Product · retention · 2026 · 15 connections
• Glossary CUPED

  CUPED cuts metric variance with a pre-experiment covariate, delivering the same effect estimate at a smaller standard error and higher power.

  Experiments · ab-testing · 2026 · 15 connections
• Glossary DAU / MAU and stickiness

  DAU and MAU are the unique active users per day and per month; their ratio, stickiness, shows how far the product became a habit.

  Product · 2026 · 10 connections
• Glossary Funnel

  A funnel is a sequence of steps from entry to a target action with conversion between them. It shows where users are lost.

  Product · 2026 · 12 connections
• Glossary LTV

  LTV is the total value of a user over their entire lifetime. It is the key benchmark for acquisition and retention economics.

  Business · segmentation · 2026 · 14 connections
• Glossary Minimum detectable effect (MDE)

  The MDE is the smallest effect a test can reliably detect given its sample size, significance level, and power.

  Statistics · ab-testing · 2026 · 11 connections
• Glossary North Star metric

  A North Star metric reflects the value the product delivers to users and predicts long-term business growth.

  Product · 2026 · 6 connections
• Glossary NPS

  NPS is a loyalty index from a single 0–10 question: the share of promoters minus the share of detractors, sensitive to how responses are collected.

  Business · segmentation · 2026 · 6 connections
• Glossary p-value

  A p-value is the probability of seeing an effect at least as large as observed when the null hypothesis is true. It is not the chance the effect is random.

  Statistics · ab-testing · 2026 · 13 connections
• Glossary Retention

  Retention is the share of users who return to the product after their first interaction. It is a core measure of value and stickiness.

  Product · retention · 2026 · 25 connections
• Glossary Retention curve

  A retention curve plots the share of returning users by period after start. Its shape is the signature of the product and its stability.

  Product · retention · 2026 · 6 connections
• Glossary RFM segmentation

  RFM segmentation splits customers by recency, frequency, and monetary value. A classic rule-based way to find valuable and lapsing segments.

  Data · segmentation · 2026 · 10 connections
• Glossary Segmentation

  Segmentation splits a base into homogeneous groups by behaviour or value, so different groups receive different treatment.

  Data · segmentation · 2026 · 9 connections
• Glossary Sample Ratio Mismatch (SRM)

  SRM is a gap between the observed and expected split. On a large sample it signals broken randomisation, not random noise.

  Experiments · ab-testing · 2026 · 13 connections
• Glossary Statistical power

  Statistical power is the probability of detecting an effect of a given size when it truly exists. It equals one minus the Type II error rate.

  Statistics · ab-testing · 2026 · 13 connections
• Glossary Time to convert

  Time to convert is the delay from entry to a target action. It shows how fast a cohort converts, not just its final share.

  Product · retention · 2026 · 7 connections
• Glossary Uplift modelling

  Uplift modelling predicts the causal effect of a treatment on an individual user rather than their outcome — which is what separates it from a plain forecast.

  Experiments · ab-testing · 2026 · 11 connections
• Projects Airflow DAG Platform

  Report platform on Apache Airflow 3: every report is a versioned DAG on a fetch → validate → write → record_metrics template, with DQ validation, run alerts and observability metrics.

  Engineering · sql · 2026 · 9 connections
• Projects This Portfolio Site

  The site you are reading: Astro 7, TypeScript and Markdown content collections, static output, dark/light theme, RSS, sitemap, JSON-LD and GitHub Pages deployment.

  Engineering · 2026 · 2 connections
• Projects Volta Neobank — Product Analytics

  Fixed a neobank's onboarding bottleneck with an A/B test: +5.72pp KYC conversion, €656K/year. 23 projects: funnel → A/B → retention → segmentation → Market & Jobs → RAT v2 → causal.

  Experiments · python · 2026 · 53 connections
• Projects SQL Analytics Case Study

  26 SQL cases: 25 on a synthetic ~183k-event dataset + 1 real-data case on UCI Online Retail II. Funnel, retention, LTV, attribution, anomalies, dbt layer on DuckDB, live report on GitHub Pages.

  Analytics · sql · 2026 · 19 connections
• Projects Causal / Uplift — CUPED and Individual Treatment Effects

  CUPED cut the standard error by 26% — the same power on 5.5k users per arm instead of 10k. At n=6000 uplift metrics stay noise-dominated: QINI and AUUC cannot separate an oracle from random.

  Experiments · python · 2026 · 17 connections
• Projects A/B Testing Methodology Toolkit

  18 A/B methodology modules, each calibrated by simulation: an A/A null check holds Type I error at α, and power curves show the effect size each method can actually detect.

  Experiments · python · 2026 · 20 connections
• Projects TaskFlow — PostHog Product Analytics Pipeline

  7 analyses from a typed event catalog: a SaaS product instrumented with PostHog end-to-end — funnel, retention, A/B, revenue/LTV and time-to-convert, not a demo on a ready-made CSV.

  Product · python · 2026 · 12 connections
• Projects Cohort Analysis Dashboard

  A triangular cohort retention and LTV matrix on synthetic data: per-cohort ARPU and LTV with an observation-age caveat, plus a Tableau-ready CSV and Hyper export.

  Analytics · python · 2026 · 13 connections
• Projects Browser Mini-Games — Analytics Arcade

  10 playable mini-games as self-contained SVGs: 7 analytics (A/B to p<0.05, funnel drop, cohort catch, retention day) and 3 arcade. One file per game, zero dependencies, no build.

  Experiments · retention · 2026 · 10 connections
• Projects Python Analytics Playground

  ≥80% pytest coverage on a module-based Python analytics toolkit: loading, cleaning, EDA and visualization assembled into one reusable pipeline — the base I copy into every new analysis.

  Analytics · python · 2026 · 10 connections
• Projects Scrolly English Speaking

  A scrollytelling guide to spoken English for workplace conversations (A2–B1): an MDX narrative with D3 visualizations, content and visuals decoupled. Built on Astro 6.

  Engineering · 2026 · 2 connections
• Projects ML Fundamentals — scikit-learn Pipeline

  A leak-free scikit-learn pipeline: missing-data imputation, categorical encoding, a deterministic split and feature scaling, all in one ColumnTransformer fit on the training split only.

  Analytics · python · 2026 · 5 connections
• Projects Product Analytics Dashboard (Streamlit)

  A product analytics dashboard on a synthetic SaaS dataset (8,000 users): AARRR funnel, cohort retention, revenue (MRR, ARPU, churn) and segmentation — 5 Streamlit pages.

  Product · python · 2026 · 12 connections
• Articles Why the aggregate lies: cohort retention triangles

  Average retention across all cohorts mixes customers with different lifetimes and lies upward. How to build a triangular cohort matrix, read observation age, and compare cohorts honestly.

  Guides · python · 2026 · 10 connections
• Articles Calibrating A/B methods with simulation: how to verify a method before production

  A p-value from a textbook doesn't prove a method — a simulation does. How to test every A/B module with an A/A check under the null and a power curve under the effect: CUPED, peeking, ratio metrics, mSPRT.

  Guides · python · 2026 · 11 connections
• Projects Sales Calls Analytics Dashboard

  A Streamlit dashboard for AI sales calls on 16,891 synthetic calls: a 4-step funnel (greeting → offer → meeting → qualification), leak breakdown by type, and transcript drill-down.

  Analytics · python · 2026 · 4 connections