Key Insights

2024 vs 2025 Performance:

• 2024 was financially worse with volatile cash flow, 5 deficit months, and a net loss

• 2025 achieved perfect balance with remarkable stability despite 18% higher transaction volume

Income & Expense Patterns:

• Top 10 creditors and debtors identified through data visualization

• Clear patterns emerged in recurring income sources and expense categories

Telecom Spending:

• Over ₦100K spent on airtime in 2025 (109% increase from 2024)

• Less than ₦50K spent on data bundles purchased directly from app

• Significant shift in telecommunication spending behavior

Recommendations and Future Improvements

1. Maintain 2025's Financial Discipline: Continue the balanced approach that eliminated deficits.

2. Monitor Airtime Spending: The 109% increase warrants attention—consider switching to bundle plans for cost savings.

3. Leverage Data Bundles: Direct app purchases are underutilized; could reduce overall telecom costs.

4. Track Monthly Cash Flow: Implement regular monitoring to avoid returning to 2024's volatility.

5. Optimize Top Expenses: Focus on the identified top debtors for potential cost reduction opportunities.

Conclusion

This analysis revealed a clear financial transformation from 2024's instability to 2025's balanced performance. Despite handling 18% more transactions, disciplined financial management eliminated deficits and achieved perfect balance. However, the dramatic increase in airtime spending presents an opportunity for optimization. For aspiring data analysts, bank statement analysis offers practical, challenging work that sharpens analytical thinking and delivers actionable insights. 

Key Insights

Overall Sentiment Distribution:

• 47.59% Negative sentiment across all outlets

• 28.89% Neutral sentiment

• 23.52% Positive sentiment

Coverage Volume:

• 13,000 total headlines analyzed across three outlets

• Legit.ng led with over 7,000 headlines, the highest volume among all sources

Sentiment by Outlet:

• All three outlets exhibited predominantly negative coverage

• Sahara Reporters recorded the highest percentage of negative headlines, reflecting its investigative focus

Temporal Trends:

• Headline coverage surged in Q4 2025, particularly in November and December

• This increase coincided with heightened negative sentiment, linked to escalating security challenges including banditry, kidnappings, terrorism, and tragic accidents

Weekly Patterns:

• Negative sentiment distribution varied across days of the week, revealing patterns in editorial focus and news cycles

Limitations

• Analysis is based on headline text only, not full article content

• Automated sentiment classification may not fully capture context, sarcasm, or nuanced language

• Data represents a snapshot of 2025 and may not reflect longer-term trends

Recommendations and Future Improvements

1. Incorporate full article text for deeper, more nuanced sentiment analysis

2. Compare multiple sentiment models (e.g., VADER, TextBlob, Transformer-based models) for validation

3. Add topic modeling to link sentiment with specific themes such as security, economy, or governance

4. Automate data refresh for real-time sentiment monitoring and continuous insights

Conclusion

This project demonstrates how sentiment analysis combined with structured data modeling and visualization delivers valuable insights into media reporting patterns. By analyzing 13,000+ headlines across multiple outlets, the dashboard provides a comprehensive, data-driven view of how Nigeria was portrayed in 2025, revealing critical trends that align with real-world events and challenges.

The goal for this project was to transform raw transactional data of the client into an interactive, insights-driven dashboard that tracks core business KPIs, identifies sales trends, and informs strategic marketing and operational decisions.

The client faced couldn't make sense of his data. The store had accumulated large volumes of sales data but lacked a centralized, visual reporting system. The marketing and management teams needed a clear view of performance across products, regions, and time to understand what was driving revenue and where opportunities for optimization existed.

As the sole analyst on this project, I owned the end-to-end process from data ingestion and modeling to visualization and insight generation. I designed a star schema in Power BI to efficiently relate sales transactions to product and factory data. I built a custom date table for time intelligence, enabling trend analysis across months and seasons.

I developed an interactive Power BI dashboard that delivers:

• At-a-glance KPIs: Total Revenue ($141.8K), Profit ($93.4K), Orders (10K), and Quantity Sold (39K).

• Trend Analysis: Monthly revenue and profit trends, highlighting December peaks and February lows.

• Product & Category Intelligence: Identification of top-performing products (Wonka Bars) and divisions (Chocolate).

• Geographical & Operational Insights: Breakdowns by country, city, factory performance, and preferred shipping methods.

Key Insights:

• Seasonality: Sales peak in December and dip in February, indicating a strong holiday effect.

• Product Dominance: Wonka Bars alone generated over $100K in revenue, outperforming all other products combined.

• Regional Performance: The United States drives the vast majority of sales, with New York City, Los Angeles, and Philadelphia as top cities.

• Operational Behavior: Standard Class shipping is the most frequently selected option by customers.

The dashboard provided a single source of truth for performance tracking. Based on the findings, I recommended:

1. Launching targeted promotions during slower months to flatten seasonal dips.

2. Increasing production and marketing focus on top sellers like Wonka Bars and the Chocolate division.

3. Exploring logistical optimizations to support peak-period demand.

4. Developing market expansion strategies for underperforming regions like Canada.

Tools & Technologies: Power BI, Power Query, DAX, Data Modeling, Star Schema Design, Dashboard Visualization.

The initial requirement was to extract data “as far back as possible” efficiently and deliver it in a clean, structured format. 

Instead of attempting to render the entire site, I conducted a deep investigation into the network layer using browser developer tools. This revealed that the site was loading data via a public JSON API. By mimicking these API requests, I bypassed the need for heavy browser automation and accessed the data directly at the source. 

I engineered a lightweight Python scraper that:

• Programmatically called the discovered API endpoints.

• Iterated through date ranges to collect several years of historical race data.

• Extracted and structured the required columns (horse names, odds, finishing positions, race times, tracks).

• Cleaned, normalized, and exported the complete dataset into a ready-to-use CSV file.

The Outcome:

✅ Delivered a complete, clean historical dataset spanning multiple years in under 24 hours.

✅ Enabled the client to immediately begin model development without data collection delays.

✅ Built a reusable and efficient data pipeline that could be scheduled for future updates.

✅ Provided documentation on the data structure and the method used for future scalability. 

Technologies Used: Scrapy, Python, API Reverse-Engineering, Data Cleaning.

To automate the collection of real-time pricing and inventory data for mobile phones from a leading Nigerian online marketplace (Jumia), enabling dynamic pricing strategy and market positioning. I developed a targeted web scraping solution focused on the mobile phone category. The scraper was designed to navigate product listings, extract key details, and handle the site's structure consistently. Special attention was given to managing product variants and ensuring data accuracy across thousands of listings. 

I built a Python scraper that:

• Systematically crawled the Jumia mobile phone category, collecting product names, current prices, discount percentages, seller information, and stock status.

• Cleaned and standardized the data (e.g., converting prices to numbers, categorizing brands).

• Structured the output into a daily CSV report and a live-updating spreadsheet for immediate client access.

• Implemented error-handling and respectful crawling delays to ensure long-term reliability.

The Outcome:

✅ Eliminated 15+ hours of manual weekly data collection, freeing the team for strategic analysis.
✅ Provided the client with a clear, daily view of competitor pricing and market trends.
✅ Enabled data-driven repricing decisions, helping to improve sales margins and campaign timing.
✅ Delivered a reliable, maintainable data pipeline that required minimal oversight.

Technologies Used: Python, Web Scraping (BeautifulSoup/Requests), Pandas, Data Cleaning, Automation