Articles on statistics, machine learning, financial data science, and the Nigerian data ecosystem. Published twice monthly, written for practitioners not beginners.
Frequentist methods dominate statistical education but Bayesian inference offers structural advantages for small-sample problems, hierarchical data, and updating beliefs with new evidence. This article explains when and why to switch frameworks.
Read ArticleBlack-box ML models are increasingly required to provide interpretable outputs for regulatory and ethical compliance. This article covers SHAP (SHapley Additive exPlanations) values with a worked XGBoost example on credit scoring data.
Read ArticleA practical walkthrough of ALM gap analysis methodology for non-banking financial companies. Covers maturity bucket construction, net interest income sensitivity, and the regulatory implications of a negative repricing gap under RBI guidelines.
Read ArticleThe phrase "correlation is not causation" is repeated everywhere but rarely followed up with concrete methods. This article introduces propensity score matching, instrumental variables, and difference-in-differences for estimating causal effects in observational data.
Read ArticleAn analysis of 340 data science job postings from Nigerian companies and multinational subsidiaries operating in Nigeria. Python, SQL, and Power BI appear in 89% of listings. Bayesian statistics and causal inference are present in under 12%.
Read ArticleBoth Weibull regression and Cox proportional hazards models are used for time-to-event data, but they make different assumptions and answer slightly different questions. This article covers model selection, interpretation, and the test for the proportional hazards assumption.
Read Article