1) Which dataset did you use?

The Adventure Works dataset — a fictional cycling equipment retailer covering sales transactions from January 2015 to July 2017. It spans six markets across North America,Europe and the Pacific, with customer demographics, product details and financial metrics all merged from multiple individual tables into a single master table.

2) How did you analyze or prepare the data?

The dataset is a single master table of 56K transactions — relatively clean but requiring significant analytical work before it could tell a coherent story. Further preparation happened on two fronts.

Inside OAC, a suite of calculated fields was built from scratch — Avg Order Value, GP margin %, revenue per customer, GP per customer and income band segmentation among others. During the data preparation phase, three purpose built datasets were created to enrich the master dataset — a category migration matrix tracking how customers moved across product categories between purchases, a product retention table measuring repeat purchase rates and second order values per product, and a purchase path frequency analysis identifying the most common multi-step buying sequences. Each was blended into OAC to extend the analytical depth beyond what the raw transaction data alone could support.

The analytical layer went further still — ML clustering across the product catalogue, statistical outlier detection across markets and time series forecasting on revenue and order trends. In total the dashboard draws on one master dataset, three purpose built analytical datasets and fourteen calculated fields — all working together to move the story from descriptive to diagnostic to predictive.

3) Who is the intended audience for your visualization?

The dashboard is designed to serve multiple audiences depending on need. Senior leadership gets the strategic narrative — market efficiency, growth drivers and concrete recommendations. Regional and product managers can use the interactive filters to drill into their specific area of responsibility — clicking a country, product or customer segment updates all visuals contextually. Analysts and operations teams can explore the underlying data through the interactive canvas elements to answer their own questions without needing a separate report. The intent was to build something that adds value at every level of the organisation — from boardroom decisions to day to day operational queries.

4) What is your visualization about and what question or problem does it address?

At its core the dashboard asks — how did Adventure Works grow, what drove that growth, and what should the business do next?

That single question breaks down into several more specific ones that each canvas addresses in turn.How has revenue and profitability trended over time, and did margin hold up as the business scaled? Which markets are generating the most revenue and which are generating the most value per customer — and why is Australia so different from the United States? Which products are driving the business and which are quietly delivering the highest margins? Who are Adventure Works' customers, how do they behave after their first purchase and who are the most valuable relationships in the portfolio?

Each canvas builds on the last — from revenue performance to regional efficiency, product strategy, customer behaviour and finally concrete recommendations. The intent was to build something that doesn't just describe what happened but gives a clear basis for deciding what to do next.

5) Did you use any Oracle Analytics AI features when building your visualization?

Yes. ML clustering on the Products canvas automatically segmented the catalogue into four distinct groups, which surfaced the accessories margin story that wouldn't have been obvious otherwise. Outlier detection on the Regional canvas flagged Australia and USA as statistically significant outliers in revenue per customer — that became one of the dashboard's strongest insights. Forecasting with confidence intervals was applied to the revenue and order volume trends to give a forward looking dimension to what is otherwise a historical dataset.

Here's the dva file:-