Connecting business data to decisions
This university capstone investigated how transaction, demographic and seasonal retail information can be prepared together to support reporting and analytical decisions. It connects data engineering, predictive modelling and business intelligence in one workflow.
Retail analysis depends on understanding transactions, customer patterns and seasonal factors across datasets. The capstone examined how multiple sources could be organised for clearer business reporting.
Retail datasets often describe different aspects of the same business. Transaction histories show purchasing activity, while demographic and seasonal information provides additional context. To interpret them together, analytical work must clarify the time periods, source fields and data definitions.
The following stages explain the workflow and the technical decisions behind this project.
The project combined transaction, demographic and seasonal datasets through extraction, transformation and loading (ETL). Cleaning and preparation were necessary before joining or comparing different kinds of information.
Python, SQL and scikit-learn supported data analysis and predictive-modelling exploration. Modelling follows the preparation stage because inconsistent inputs can undermine the usefulness of predictions.
Power BI and Tableau were used to present results in visual form. The aim was to make analytical patterns and relationships more accessible for business interpretation.
The final analytical workflow brought data preparation, modelling and visual reporting together, illustrating how a question can move from source records to decision-support views.
Prepared analytical datasets, explored predictive models and developed Power BI/Tableau reporting views.
Reporting is most useful when business questions and data definitions are clear before selecting charts or models.
Data integration, information quality, reporting requirements and the design of useful decision-support outputs.
The available résumé does not supply the underlying dataset, screenshots, benchmark metrics or independently verified commercial outcomes. This is an academic capstone, not a production retail platform.