Power analysis: effect size, significance level, power, test type, design-adjusted sample size computation with attrition corrections.
A study with too few participants cannot detect a real effect: it is underpowered. A study with too many wastes resources and detects trivial effects ...
Propensity score matching: logistic regression, nearest-neighbor matching, balance diagnostic, causal effect estimation with sensitivity.
Correlation is not causation, but sometimes you cannot run a randomized experiment. Propensity score matching (PSM) estimates causal effects from obse...
Multi-arm bandit with Thompson Sampling: Beta-Bernoulli posteriors, regret minimization, convergence criteria.
A/B tests have a fixed hypothesis and duration. Multi-arm bandits continuously allocate traffic to better-performing variants, minimizing regret while...
Funnel drop-off analysis: step-wise enumeration, leak prioritization with downstream impact, effort-impact matrix.
Aggregate conversion hides which step loses the most users. Fixing the step with the highest absolute drop-off is often wrong - fix the step with high...
A/B test design: sample size formula, MDE, duration rules, no-peeking sequential testing checklist.
Half of all A/B tests fail because they run too short. 'Run for a week or until significant' biases toward false positives and false negatives. SAMPL...
Time series anomaly detection: STL decomposition, dynamic MAD thresholds, multi-seasonality, alerting rules.
Most anomaly detection runs a static threshold ("alert if error rate >5%"). This fails on seasonal data: 5% errors at 3 AM is different from 5% at 3 P...
Model interpretation: permutation importance, partial dependence plots, SHAP values, correlated feature handling.
Feature importance tells you which inputs drive your model's predictions. Tree-based models give built-in importance, but it can be misleading (favors...
Cohort retention analysis: curve types, churn segmentation (power/core/casual), Kaplan-Meier survival times.
Aggregate metrics lie. A 95% retention rate sounds healthy until you see that month-6 retention is 40% and the 'average' is pulled up by week-1 users ...
Analytical framework for Analytics.
Apply structured reasoning to this domain. Decompose, analyze, and produce actionable output.
Bayesian A/B analysis: priors, posteriors, win probability, practical significance, expected loss.
Frequentist p-values tell you whether a difference exists, not which variant is better or by how much. Bayesian analysis gives the probability B beats...
Expert-level analyse prompt with detailed methodology and actionable recommendations.
You are a Senior Data Scientist with 12+ years of experience in statistical modeling, machine learning, and data-driven decision making. You hold a Ph...