skills

Suggestions for scoping and writing an Analytics Project Brief — the one-page artifact that defines problem, metrics, counter-metrics, stakeholders, methodology, success criteria, and pre-mortem before any analysis begins. Use when the task is to draft, review, or critique a project brief, scope an analytics project, define KPIs, identify counter-metrics or blockers, or prepare a stakeholder map. Not for technical implementation — see ml-modeling, statistical-modeling, or data-warehousing for that.

Best-practice suggestions for predictive ML modelling with scikit-learn — Pipelines, ColumnTransformer, cross-validation, hyperparameter search, honest train/test evaluation, model leaderboards, and threshold/loss-based decisions. Use when building, comparing or reviewing supervised ML models (regression or classification) in Python notebooks or scripts.

Best-practice suggestions for statistical / inferential modelling in Python — OLS and logistic regression with statsmodels, robust standard errors, side-by-side regression tables with stargazer, confidence and prediction intervals, hypothesis tests, and significance reporting. Use when interpreting coefficients, building explanatory regressions, comparing nested models, reporting confidence/prediction intervals, or testing whether an effect is statistically significant.

Best-practice suggestions for time series exploration and forecasting in Python — datetime indexing, resampling, temporal train/test splits, decomposition, ACF/PACF, stationarity checks, ARIMA/SARIMA/SARIMAX, AutoGluon TimeSeriesPredictor, backtesting, forecast metrics, and prediction intervals. Use when analyzing, building, comparing, or reviewing forecasts for dated/ordered data such as demand, energy, sales, traffic, sensors, macro, or finance series.

Best practices for designing data warehouses and analytical pipelines using the bronze/silver/gold medallion architecture, validations-as-code, and idempotent transforms. Use when building or modifying data pipelines, ETL/ELT jobs, dbt models, SQL warehouses, lakehouses, or any layered analytics workload (DuckDB, Snowflake, BigQuery, Postgres, Spark, etc.).

Technical setup skill for analytics and data science projects — repository scaffolding, folder structure (dev/prod split, data layers, numbered notebooks), environment management (uv, venv, dotenv), pre-commit hooks for notebook output clearing, branching and commit conventions, .gitignore patterns, AGENTS.md creation, database/storage I/O patterns, and production orchestration notebooks. Use when initialising a new analytics project, setting up a repo for a data science team, or creating an AGENTS.md file.

AI-powered code review using CodeRabbit. Default code-review skill. Trigger for any explicit review request AND autonomously when the agent thinks a review is needed (code/PR/quality/security).