ChiedzaRene/WTB-Capstone-Project-Group-18-Smart-Analytics-For-SMEs-In-Nigeria-

Women Techsters Bootcamp Data Analysis Capstone Project # SME Capstone Project - AI-Powered Business Intelligence Dashboard **Group 18 · Tech4dev 5.1** Smart analytics for SMEs in Nigeria, built on Power BI · Data Analysis & Technical Project Management tracks --- ## Table of Contents - Overview - Business Context - Data Pipeline - Dashboard Structure - Key Figures - Key Insights - Recommendations - Tech Stack - Team --- ## Overview This project delivers a full business intelligence solution for **Inner Space Interior Design Company**, a Nigerian SME offering design, renovation, and fit-out services. Starting from a raw, messy project-level dataset (300 records), the team cleaned and structured the data, then built an interactive Power BI dashboard to turn it into decisions a manager can actually act on revenue performance, service profitability, pipeline health, and customer/regional insights. The goal wasn't just to visualize numbers, but to answer the questions an SME owner genuinely needs answered: - Are we making enough profit per project? - Are we finishing projects on time and on budget? - Which clients, states, or services bring in the most value? - Where is the business losing money or leads? ## Business Context Inner Space Interior Design Company runs project-based work rather than selling a single product — each project has its own client, budget, team, and timeline, spanning both residential and commercial work across multiple Nigerian states. ## Data Pipeline 1. **Raw data audit** - 300 project records with missing values, inconsistent text formatting, duplicate entries, and logical inconsistencies (e.g. completion dates on projects that were still ongoing). 2. **Cleaning** - handled column-by-column based on what each gap actually meant: - Derived values from related fields where possible (e.g. Completion Date from Start Date + Days to Complete) - Matched repeat clients via Client ID to recover missing names - Flagged rather than fabricated values that couldn't be recovered (e.g. "Unknown", "No …