Focus Track
Data Science & Analytics
From descriptive dashboards to prescriptive actions, grounded in South Africa’s data realities.
Descriptive
What happened? Counts, totals, distributions.
Diagnostic
Why did it happen? Segment, correlate, drill-down.
Predictive
What might happen? Forecast demand, risk scores.
Prescriptive
What should we do? Recommendations and actions.
South African Contexts
Education analytics: module pass rates, tutoring impact, fairness checks.
Public health reporting and dashboards for clinics and campuses.
Transport and mobility: class schedule vs. shuttle usage, travel times.
Energy and utilities: residence consumption, demand response signals.
Retail and campus stores: basket analysis, stock and supply planning.
Tools
- • SQL fundamentals; joins, window functions, and CTEs.
- • Python with pandas, notebooks for exploration.
- • Visualization: Power BI/Tableau, lightweight web dashboards.
- • Transformation: dbt-style modeling; tests and documentation.
- • Warehouses and lakes: BigQuery/Snowflake concepts; Parquet/Delta basics.
Workflow
- • Ask clear questions with stakeholders before pulling data.
- • Prototype quickly in notebooks; promote to versioned models.
- • Add tests: row counts, freshness, null thresholds.
- • Publish simple, readable visuals with context and caveats.
Data Quality in the Real World
Missing data handling: imputation vs. flagging; track fill rates.
Sampling: know who is represented; avoid urban-only or single-language bias.
Privacy: minimise personal identifiers; aggregate where possible.
Lineage: document source systems, refresh cadence, and owners.
Validation: freshness checks, anomaly detection, reconciliation.