EKE
DEBORAH
DATA ANALYST
Hi, I'm Eke Deborah, a Data Analyst based in Lagos. I turn complex datasets into clear dashboards that help organizations make faster, smarter decisions using Excel, SQL, Power BI, and Python.
View CV 👁
Hi, I'm Eke Deborah, a Data Analyst based in Lagos. I turn complex datasets into clear dashboards that help organizations make faster, smarter decisions using Excel, SQL, Power BI, and Python.
View CV 👁Check out some of the case studies I made as a Data Analyst
End-to-end analysis of US trucking operations using Power BI + SQL. Cleaned 14 tables with 550K+ rows, built a star schema, and delivered insights that cut delivery delays and $1.6M in preventable incidents.
Technologies: SQL, Power BI, Power Query, DAX
VIEW PROJECT ON GITHUB
Analyzed e-commerce sales data to track revenue performance and customer behavior patterns. Used SQL to clean and extract data, then built an interactive Power BI dashboard. Identified top performing products and referral sources.
Technologies: SQL, Power BI
VIEW PROJECT
Cleaned and organized sales data in Excel, then built an interactive dashboard with Pivot Tables and slicers. Tracked KPIs, sales trends, and regional performance for data-driven decisions.
VIEW PROJECTThis project delivers a fully interactive 4-page Power BI dashboard analysing the end-to-end operations of a US-based trucking and logistics company. The analysis covers 85,410 loads, 150 active drivers, 200 customers, and 122 million miles of freight movement across a three-year period from 2022 to 2024.
Trucking operations generate data across multiple disconnected systems — loads, trips, fuel purchases, maintenance records, delivery events, and safety incidents. Without a unified analytical view, operations managers cannot see which routes are profitable, which drivers are underperforming, where safety incidents are concentrated, or which customers are generating the most value. Decisions get made on intuition rather than evidence.
With $298.6 million in total revenue and only a 55.67% on-time delivery rate, this business is generating strong top-line numbers while quietly experiencing a service quality crisis. Nearly half of all 85,410 deliveries arrived late. The dashboard surfaces these issues and provides the granularity needed to act on them.
To connect 14 tables of operational logistics data into a single Power BI model, build a suite of DAX measures covering revenue, profitability, safety, and service quality, and deliver a four-page interactive dashboard that answers the most critical business questions across operations, driver management, fleet safety, and customer analysis.
The executive summary page showing headline KPIs, top 10 customers by revenue, customer type revenue split, monthly incident trend, fleet utilization gauge, vehicle damage cost by incident type, average revenue per load, and a monthly financial summary table.
Covers on-time delivery rate, late delivery count, active driver count, profit per driver, booking type split (Dedicated / Contract / Spot), total loads delivered by driver, driver experience vs incident rate, and route profit by lane.
Tracks total damage cost, preventable incident rate, total downtime hours, total maintenance cost, maintenance cost by type, incidents by home terminal, incident type breakdown, and monthly incident trend.
Analyses total customers, total loads delivered, total revenue, late deliveries by customer, load volume and revenue ranking by customer name, and monthly profit trend.
| Total Tables | 14 connected tables |
| Total Loads | 85,410 |
| Total Customers | 200 |
| Active Drivers | 150 |
| Total Miles Driven | 122,159,201 |
| Period Covered | January 2022 – December 2024 |
| Model Type | Star schema with central Date table |
| Dimensions | customers, drivers, trucks, trailers, routes, facilities |
| Facts | loads, trips, fuel_purchases, maintenance_records, delivery_events, safety_incidents |
| Aggregates | driver_monthly_metrics, truck_utilization_metrics |
| Power BI Desktop | Data modelling, DAX measures, and dashboard development |
| Power Query (M) | Data cleaning, type standardisation, and transformation |
| DAX | Custom KPIs, calculated measures, and time intelligence |
| Star Schema | Relationship modelling across 14 tables |
| SQL | Source data validation and cross-table join queries |
| Excel / CSV | Source data format |
All cleaning was performed in Power Query before the data entered the Power BI model.
The full data model was built in Power BI Model view using a star schema pattern. All 14 tables are connected through clearly defined one-to-many relationships with single-direction filtering.
Average MPG
Total Miles Driven
Total Incidents
Fuel Cost per Lane
Total Downtime Hour
Total Maintenance Cost
Total Load Delivered
Revenue % by Customer Type
Profit Margin %
Profit per Driver
Late Deliveries
Active Drivers
On-time-delivery %
Total Customers
| Measure | Table | Format | Formula |
|---|---|---|---|
| AVG MPG | driver_monthly_metrics | General | AVERAGE(driver_monthly_metrics[average_mpg]) |
| Total Miles Driven | trips | Whole Number | SUM(trips[actual_distance_miles]) |
| TOTAL INCIDENTS | safety_incidents | Whole Number | COUNTROWS(safety_incidents) |
| Total Load Delivered | loads | Decimal Number | COUNTROWS(loads) |
| Profit Margin% | loads | Percentage | DIVIDE(routes[Route Profit], [Total Revenue]) |
| On-time-delivery % | driver_monthly_metrics | Percentage (2dp) | DIVIDE(COUNTROWS(FILTER(delivery_events, delivery_events[on_time_flag]=TRUE)), COUNTROWS(delivery_events)) |
| Total Revenue | $298.6M |
| Total Profit | $197.3M |
| Total Miles Driven | 122M |
| Average MPG | 6.50 |
| Profit Margin | 66.07% |
| Fleet Utilisation Rate | 83% |
On-Time Delivery Rate: 55.67% | Late Deliveries: 47,308 | Active Drivers: 150 | Top Lane by Profit: NC → OR ($3.9M)
Total Damage Cost: $1.60M | Preventable Incident Rate: 37.65% | Total Downtime Hours: 72,230
Total Customers: 200 | Total Loads Delivered: 85,410 | Top Customer by Revenue: First Group ($10.4M)
14 Raw Source Tables (CSV / Excel)
│
▼
Data Cleaning in Power Query
├── Remove duplicates
├── Fix data types
├── Standardise text columns
├── Handle missing values
└── Build central Date table
│
▼
Data Modelling (Star Schema)
├── Define relationships (1-to-many)
├── Set filter directions (single)
└── Connect Date table to all facts
│
▼
DAX Measures
├── Revenue, Profit, Margin %
├── On-Time Delivery %
├── Late Deliveries count
├── Fleet Utilisation Rate
├── Preventable Incident Rate
└── Cost per Mile
│
▼
4-Page Interactive Dashboard
├── Page 1: Overview
├── Page 2: Drivers & Route Performance
├── Page 3: Maintenance & Safety
└── Page 4: Customers Analysis
│
▼
Key Insights & Business Recommendations
Logistics-Operations-Dashboard/ │ ├── Data/ │ ├── customers.csv │ ├── drivers.csv │ ├── trucks.csv │ ├── trailers.csv │ ├── routes.csv │ ├── facilities.csv │ ├── loads.csv │ ├── trips.csv │ ├── fuel_purchases.csv │ ├── maintenance_records.csv │ ├── delivery_events.csv │ ├── safety_incidents.csv │ ├── driver_monthly_metrics.csv │ └── truck_utilization_metrics.csv │ ├── Dashboard/ │ └── Logistics_Operations_Dashboard.pbix │ ├── SQL/ │ └── validation_queries.sql │ ├── Images/ │ ├── LOGIS.png │ ├── LOGIS2.png │ ├── LOGIS3.png │ ├── LOGIS1.png │ ├── model_view.png │ └── [all dax_xxx.png files] │ └── README.md
| Skill | Evidence |
|---|---|
| Data Modelling | 14-table star schema with one-to-many relationships |
| Power Query | Cleaning, type fixing, deduplication, and Date table creation |
| DAX Measures | Revenue, margin, OTD%, cost per mile, preventable incident rate |
| Time Intelligence | Central Date table with year, quarter, month slicers |
| Dashboard Design | 4-page interactive report with consistent green/white design system |
This project transforms 14 connected logistics tables — covering 85,410 loads, 150 drivers, 200 customers, and 122 million miles of operations — into a four-page interactive Power BI dashboard that gives logistics managers complete visibility into their business. The analysis uncovered a 55.67% on-time delivery rate against an industry standard above 90%, identified the NC → OR lane as the most profitable at $3.9M, found that 37.65% of safety incidents were preventable, and revealed that Indianapolis and Las Vegas terminals carry disproportionate incident risk. Total revenue of $298.6M is strong, but the service quality picture demands operational attention before customer contracts come up for renewal. The dashboard equips decision-makers to address these issues with evidence rather than instinct.
This logistics operations dashboard demonstrates what becomes possible when operational data is treated as a strategic asset rather than a by-product of daily activity. By connecting 14 tables, building a clean star schema, and developing targeted DAX measures, the project delivers a tool that gives logistics leadership visibility into every dimension of the operation. The central finding — that more than 44% of deliveries arrive late despite strong revenue performance — is the kind of insight that only surfaces when operational data is connected and examined as a whole.
Eke Deborah — Data Analyst | Power BI · SQL · Python · Excel
🔗 LinkedIn | 📧 your.email@gmail.com | 🐙 GitHub
*Built as part of a data analytics portfolio. Dataset covers a simulated US trucking operation spanning 2022–2024.*
An end-to-end analysis of an online retail operation — cleaned and calculated in SQL, then visualized in Power BI
Interact with the dashboard: View on Power BI
Performance dashboard showing sales KPIs, order status distribution, revenue by product, and trends over time
This study analyzes a dataset of 1,200 e-commerce orders in order to understand how the business is performing across sales, order fulfillment, customers, and marketing. An online store generates a steady stream of transaction records, but on their own those records say very little. The goal of this project was to turn that raw order log into a clear picture of what is working in the business and, more importantly, what is not.
The dataset is a single table of online orders, with each row representing one order placed on the store. It captures the full lifecycle of a transaction including order ID, customer ID, order date, product, quantity, unit price, total price, shipping address, payment method, order status, tracking number, cart items, referral source, and coupon code. In total the dataset spans orders placed between January 2023 and June 2025.
For this project, the data was cleaned and prepared using SQL before being brought into Power BI for visualization. Doing the cleaning and the calculations in SQL meant the numbers were verified at the source, so that Power BI only had to focus on presenting them. The cleaning involved several steps:
1. Removed duplicate records. The order table was checked for repeated rows, and the order identifier was checked to confirm that every order appeared only once. Removing duplicates is important because a duplicated order would be counted twice and would quietly inflate both the order count and the total revenue.
2. Fixed data types and formats. Each column was reviewed to make sure it was stored in the correct form, so that dates were treated as proper dates, the quantity and item counts were treated as whole numbers, and the unit price and total price were treated as decimal values. Correct formatting matters because it determines whether dates sort properly and whether the revenue calculations return accurate results.
3. Standardized text values. Fields such as the product name, payment method, order status, and referral source were cleaned of extra spaces and made consistent in their capitalization, so that the same value would not appear in two slightly different forms and be counted as two separate categories.
4. Handled the coupon code field. Around a quarter of all orders had no coupon code at all, because those customers simply did not use a promotion. Since the field was largely empty and did not contribute meaningfully to the analysis, the coupon code column was removed so that the focus could stay on the fields that actually drive insight.
5. Calculated core measures in SQL. With the data cleaned, SQL was used to calculate the core business measures directly in the database. Total revenue was calculated, along with the average order value, the number of unique customers, the revenue generated by each product, and the totals for each year. Calculating these figures in SQL first meant they could be checked and trusted before they ever reached the dashboard.
The performance dashboard was designed to give an overview of the whole business at a glance and then let the user filter into the details. Each visual was chosen for the kind of question it answers best.
KPI cards (Total Revenue, Total Orders, Total Customers, Average Order Value). Cards are the fastest way to communicate a single important number, and together these four immediately establish the size of the business before the user examines any chart. Placing total customers directly beside total orders was a deliberate choice, because seeing that the two numbers are almost identical instantly raises an important question about whether customers are returning.
Total Orders by Order Status — column chart. A column chart compares categories side by side, which makes it easy to see how orders are distributed across delivered, shipped, pending, cancelled, and returned. This visual turned out to be the most revealing on the page, because it exposes how few orders actually reach the customer.
Total Revenue by Product — horizontal bar chart. A horizontal bar chart ranks the products clearly and leaves room for the product names to remain readable, so it is easy to see which products contribute the most and how close they are to one another.
Total Orders by Referral Source — column chart. A column chart makes it simple to compare the marketing channels against each other and to see whether any single channel dominates or whether traffic is spread evenly across them.
Total Revenue by Year — line chart. A line chart is the clearest way to display change over time, and plotting revenue by year immediately shows the direction the business is heading. It is worth noting that the data for 2025 only covers the first half of the year, so that final point represents a partial year rather than a full one.
Slicers (Payment Method, Quarter, Month). Slicers were placed at the top of the dashboard so that the entire page can be filtered by how customers paid and by time period. Selecting any of these updates every visual on the page at once, allowing the user to explore the data from different angles.
1. Order fulfillment is the most serious problem. Only about 19% of orders were actually delivered, while the remaining orders were spread across pending, shipped, cancelled, and returned. With roughly a fifth of orders ending in cancellation and another fifth in returns, far more orders fail to complete than succeed, which is well below what a healthy online store should expect.
2. There is an almost complete absence of repeat customers. There were 1,189 unique customers across 1,200 orders, which means that on average each customer placed only a single order and almost never returned. This points to a customer retention problem rather than simply a problem of attracting new buyers.
3. Revenue is trending downward. Revenue fell from about $0.55 million in 2023 to roughly $0.48 million in 2024. The figure for 2025 is lower again, but because that year only includes the first six months of data, it cannot be compared directly to the full years before it. Even so, the year-over-year decline from 2023 to 2024 is real and worth attention.
4. Revenue and marketing are balanced across categories. Revenue is fairly balanced rather than concentrated. The leading products, including the chair and the printer, each generated around $196,000, with the remaining products following closely behind. Marketing performance is similarly even, with channels such as Instagram and email leading but no single referral source standing out dramatically from the rest.
The business should focus first on its fulfillment problem, because driving more sales will not help a store that cannot deliver the orders it already has. Understanding why so many orders are cancelled or returned, whether the cause is stock issues, slow shipping, or mismatched customer expectations, should be the immediate priority.
Alongside this, the business should build a customer retention strategy, since converting even a small share of one-time buyers into repeat customers would have a major effect on revenue given how many one-time customers already exist. The high return rate in particular deserves investigation, as returns are costly and often signal a gap between what customers expect and what they receive.
A comprehensive Excel-based visualization analyzing sales, profit, and key metrics for a multi-region retail company
View the file: Download Excel
Excel dashboard showing revenue, profit, and performance metrics across products, regions, and sales reps
Before any analysis, the raw sales data was cleaned and prepared in Microsoft Excel using Power Query. The following steps were applied:
1. Removed duplicate records. The dataset was checked for repeated rows so that no sale would be counted twice and totals would not be inflated.
2. Fixed data types. Dates, revenue, COGS, and profit columns were set to their correct formats so calculations and charts would behave properly.
3. Standardized text values. Extra spaces were trimmed and capitalization of product names, cities, and regions was unified so the same value would not appear as two separate categories.
4. Created calculated fields. Profit, COGS, and profit margin measures were built to measure profitability consistently across every product, sales rep, city, and region.
5. Built PivotTables and PivotCharts. The clean data was summarized and interactive slicers were connected for region and product category to enable dynamic filtering.
The dashboard was built using pivot tables and charts to answer key business questions at a glance. Each visual was chosen for the type of comparison it communicates best.
Profit by Product — bar chart. A bar chart ranks products clearly to show which items drive the most profit. Laptop A13 is the most profitable product, generating ₦105,336,000 — far ahead of every other item. Sofa Classic at ₦69,248,000 and Desktop PC D21 at ₦68,640,000 follow as strong performers. Blender B10 at ₦6,870,000 and Microwave M20 at ₦13,416,000 are the lowest contributors, highlighting potential areas for pricing or promotion review.
Sales Representative Performance — column chart. A column chart compares revenue side by side across the team. Peter leads with ₦434,805,000 in revenue. Musa at ₦397,665,000 and David at ₦378,660,000 follow closely, with the team tightly grouped — a sign of a consistent, capable sales force rather than reliance on a single performer.
Cost by City — column chart. A column chart shows COGS distribution across locations. Lagos carries the highest cost of goods sold at ₦493,812,000, followed by Port Harcourt at ₦486,788,000 and Abuja at ₦458,784,000. COGS is spread fairly evenly across the cities, indicating broad operations rather than dependence on one market.
Customer Trend — line chart. A line chart displays change over time. The business serves 2,098 customers, but the monthly trend shows a sharp decline toward the end of the period, falling from around 500 customers to a small handful. This steep drop is the most urgent signal in the data and warrants immediate investigation.
1. Product performance varies significantly. Focus stock and marketing on the clear profit leaders — Laptop A13, Sofa Classic, and Desktop PC D21. Review the pricing or sourcing of low-profit products such as Blender B10 and Microwave M20, or replace them with higher-margin alternatives.
2. Margin improvement opportunity exists. With COGS consuming roughly 80% of revenue, even a small reduction in the cost of goods would meaningfully raise profit. Negotiate better supplier terms, particularly in high-COGS cities like Lagos and Port Harcourt.
3. Sales team performance is consistent. The sales team performs consistently. Study what Peter and Musa do well and share those practices across the team to lift overall revenue.
4. Customer retention is urgent. The sharp drop in monthly customers is the single biggest concern. Investigate the cause behind the decline and put retention measures — follow-ups, loyalty offers, and re-engagement campaigns — in place to reverse it.
I have successfully completed the Data Analytics Essentials program with Cisco Networking Academy. I gained hands-on skills in data analysis, SQL, Python, and building dashboards to turn raw data into actionable business insights.
View certificate →I have successfully completed the program with She Leads Africa. I gained practical knowledge in data fundamentals, data cleaning, basic analysis, and working with datasets to extract meaningful insights.
View certificate →
I'm open to data analyst roles and collaborations. The fastest way to reach me is email or LinkedIn.