Imran Agunbiade

Othello Institute of Technology

Lagos State University

Numbers don't lie — but they do hide. This project was about pulling back the curtain on a retail company's multi-year sales data to find what was actually driving revenue, profit, and growth. Which product categories were carrying the business? Which regions were underperforming quietly? When did sales spike, and why? I built a full visual dashboard in Python that answers all of it — clean charts, clear insights, and the kind of analysis that turns raw spreadsheet data into business decisions.

An interactive Excel dashboard analysing H1 2025 sales performance for HP Nigeria across 4 regions, 6 products, and 6 salespersons. Features 5 pivot-driven visuals, 3 cross-filtering slicers, and a monthly trend annotation flagging a 79% sales decline from January to June 2025. Designed to HP brand guidelines using a consistent blue and white colour theme.

A 3-page interactive Power BI dashboard tracking course enrollments and revenue across 19 courses in 5 categories, built around Literate Nigeria — a Lagos-based non-profit providing digital skills education to young Nigerians. Features a star schema data model across 3 related tables, 4 custom DAX measures including Month-over-Month Revenue Growth %, and cross-filtering slicers for dynamic analysis by date range and course category. Dataset is simulated, modelled after Literate Nigeria's actual course offerings.

Analysed a 700-transaction financial dataset covering five business segments, five countries, and six products across a 16-month period. Loaded raw Excel data into a PostgreSQL database using Python and SQLAlchemy, then conducted a structured SQL investigation — moving from broad totals down to segment-discount interactions, product-level unit economics, and month-on-month trends. Findings were visualised in six Matplotlib charts and synthesised into a formal findings and recommendations section. Key findings included the Enterprise segment operating at a -3.13% profit margin driven entirely by its discount structure, Channel Partners achieving 73.13% margin with strong pricing resilience, Amarilla leading on profit per unit despite low volume, the USA generating the highest revenue but one of the lowest margins, and a consistent profit decline every November across both years.

What can a 100-year-old shipwreck tell us about human behaviour and survival? More than you'd expect. I dug into the Titanic passenger dataset — 891 records covering age, gender, class, and fate — and let the data speak. After cleaning and exploring the data, patterns emerged that were equal parts fascinating and sobering: your gender, your ticket class, even what you paid determined your odds of making it out alive. I then built a Random Forest machine learning model that predicts survival outcomes with 82% accuracy. Equal parts history lesson and data science project.