Data & BI Portfolio

A retail chain trialled a layout change in three stores and needed to know if it actually worked — not just whether sales went up, since sales go up and down for a hundred reasons that have nothing to do with a layout.

• Matched trial stores 77, 86, and 88 to control stores 233, 155, and 237 using two custom functions — calculateCorrelation and calculateMagnitudeDistance — to find controls that actually tracked the trial stores' pre-trial behavior.
• Built in R with data.table for the joins and aggregations and ggplot2 for the comparison visuals, run through setDT() before any table operation.
• Delivered as an executive summary plus a full findings document, submitted as part of the Quantium job simulation.

Built HR analytics dashboards in both Power BI and Tableau — deliberately not the same dataset reshaped twice. Each uses a different underlying schema, built and reasoned through independently, because presenting them as one unified model would be dishonest about what they actually are.

R for the Quantium question, because a causal claim needs statistical rigor and a reproducible script, not a drag-and-drop chart. Power BI and Tableau for the HR dashboards, because that's stakeholder-facing work — the deliverable is a decision someone else can make without reading code.