Data Science & Analytics · Beginner Projects · Kenya
Data Analyst Projects for Beginners in Kenya: 7 Step-by-Step Projects
Ask → Check → Analyse → Explain
Build seven practical Data Analyst projects with Excel, SQL, Power BI, Python and real Kenyan public data. Each project shows what to analyse, what to check and what evidence to finish with.
Quick answer
Which Data Analyst projects should a beginner build?
Start with one small project and finish it properly. A strong beginner sequence is Excel → SQL → Power BI → Python/data quality → one end-to-end capstone. The goal is not to collect dashboards; it is to show that you can move from a clear question to a checked, defensible answer.
- Begin with Excel if this is your first analytical project.
- Add SQL and Power BI to show database and reporting skills.
- Use Kenyan public data to demonstrate sourcing, interpretation and responsible evidence.
- Finish with a capstone that combines several tools without hiding the reasoning.
The 7 projects at a glance
You do not need to complete all seven immediately. Choose a project that adds a skill you cannot yet demonstrate, complete it, then move to the next one.
| # | Project | Main tool | Level | Finish with |
|---|---|---|---|---|
| 1 | Retail sales and gross-profit analysis | Excel | Beginner | Workbook + PivotTables + report |
| 2 | Customer and order analysis | SQL | Beginner | SQL script + result tables |
| 3 | Branch-performance report | Power BI | Beginner–Intermediate | Data model + Power BI report |
| 4 | Kenya inflation analysis | Excel / Python / Power BI | Beginner–Intermediate | Official-data trend analysis |
| 5 | Data-quality audit | Python / Excel | Intermediate | Audit report + cleaned dataset |
| 6 | Kenya public-procurement analysis | SQL / Excel / Power BI | Intermediate | Public-data case study |
| 7 | End-to-end business capstone | Multiple tools | Intermediate | Complete portfolio case study |
On a small screen, swipe the table sideways to view every column.
Use one analytical habit in every project
The software changes, but the reasoning should remain disciplined. Use this sequence whenever you work with a new dataset.
Define the question
Start with a question a person could act on—not simply “analyse this CSV.”
Understand the grain
Know what one row represents before you group, join or count anything.
Inspect and clean
Check missing values, duplicates, dates, categories, types and unusual records before analysis.
Analyse and verify
Use the simplest suitable method, then independently check important totals or outputs.
Explain the evidence
Write the finding in ordinary language and distinguish what the data shows from what you infer.
State the limit
Say what the available data cannot establish and what should be investigated next.
Why this matters
Current Kenyan entry-level analytical work can involve questions from sales, operations and finance, plus SQL, Python/pandas, visualisation and careful checking of results before they are shared. Academic or personal projects can also provide evidence of hands-on experience for an early-career applicant.[1][2]
Project 1: Analyse retail sales in Excel
Beginner · Excel — Turn transaction rows into revenue, gross-profit and branch insights.
Imagine a small retailer exports the following fictional records. Your question is: Which products and branches are contributing most to revenue and gross profit?
| Date | Branch | Product | Category | Quantity | Unit price | Unit cost |
|---|---|---|---|---|---|---|
| 03/01/2026 | Migori | Product A | Household | 3 | KSh 1,200 | KSh 850 |
| 03/01/2026 | Kisii | Product B | Electronics | 1 | KSh 4,500 | KSh 3,600 |
| 04/01/2026 | Migori | Product C | Household | 5 | KSh 650 | KSh 400 |
Training data: fictional example created for this guide.
Inspect the records
Check dates, numeric quantities, consistent branch names, missing prices, duplicate rows and unusual negative quantities.
Calculate revenue
Use Revenue = Quantity × Unit Price. Product A: 3 × KSh 1,200 = KSh 3,600.
Calculate total cost
Use Total Cost = Quantity × Unit Cost. Product A: 3 × KSh 850 = KSh 2,550.
Calculate gross profit
Use Gross Profit = Revenue − Total Cost. Product A: KSh 3,600 − KSh 2,550 = KSh 1,050.
Build PivotTables
Compare branch revenue, branch gross profit, category units, monthly performance and products with strong sales but weaker margins.
Present the result
Show only the measures and charts needed to answer the original question, then write three findings in plain language.
Finish with
The original data, a cleaned analysis table, formulas, PivotTables, a small report, three written findings and at least one limitation. Example: “Migori recorded higher revenue in the selected period, while Kisii produced the stronger gross-profit percentage.”
Project 2: Analyse customer orders with SQL
Beginner · SQL — Learn to retrieve and combine information stored across related tables.
SQL remains a practical skill in current entry-level analytical work in Kenya.[1][2] Build four fictional tables: customers, orders, order_items and products. Ask: Which counties generated the highest recorded sales revenue?
Understand the grain before you join
One row in orders represents one order. One row in order_items represents one product line within an order. If an order contains four products, it can appear once in orders but four times in order_items. Misunderstanding this can create double-counting.
PostgreSQL documents querying, joins and aggregate functions as core SQL operations.[3] A query that runs is not automatically a trustworthy answer: check whether cancelled orders are included, whether unit price means the actual selling price, whether order items are duplicated and how missing counties are handled.
Repeat customers
Count orders by customer and identify customers with more than one valid order.
Category revenue
Join products to order items and compare revenue by category.
Monthly movement
Group orders by month and compare totals over time.
Frequency vs value
Find products that appear often in orders but contribute relatively little revenue.
Finish with
A small schema diagram, one SQL file containing the important queries, result tables, the checks you performed and written findings. Do not publish dozens of nearly identical queries simply to make the project look large.
Project 3: Build a Power BI branch-performance report
Beginner–Intermediate · Power BI — Practise data modelling, measures and decision-focused reporting.
A technical distinction matters: a multi-page artefact created in Power BI Desktop is a report. Microsoft describes a Power BI dashboard as a single-page canvas in the Power BI service, and dashboards are not created in Power BI Desktop.[4]
Create Sales, Branch, Product and Date tables. Ask: Which branches are performing well, and where should management investigate further?
Inspect in Power Query
Check dates, data types, duplicate sales rows, blank keys and inconsistent product categories.
Model the data
Treat Sales as the fact table and Branch, Product and Date as dimensions describing those sales events.
Build relationships
Use clear one-to-many relationships such as Date → Sales, Product → Sales and Branch → Sales.
Create measures
Build Total Revenue, Total Cost, Gross Profit, Gross Margin %, Sales Target and Target Achievement %.
Build the report
Answer branch revenue, margin, target, product and time questions with a small number of purposeful visuals.
Interpret carefully
A branch below target is a finding; it is not automatic proof of poor staff performance. Check stock, opening days, target design, demand and data completeness.
Microsoft’s modelling guidance explains that dimension tables support filtering and grouping, fact tables hold observations to summarise, and fact tables should maintain a consistent grain.[5]
Finish with
A model-view screenshot, the important measures, report screenshots, two or three findings, one data-quality issue and the limitations of the analysis.
Project 4: Analyse Kenya’s inflation data
Beginner–Intermediate · Official Kenyan data — Practise time-series analysis with an authoritative source.
The Kenya National Bureau of Statistics publishes official Consumer Price Index and inflation releases.[6] Ask: How has Kenya’s Consumer Price Index changed over the selected period?
For September 2026, KNBS reported annual inflation of 6.8%. Use the official release and its underlying series to build a longer time sequence rather than analysing one headline number in isolation.[6]
Collect the series
Record the month, CPI, monthly inflation and annual inflation, together with source, release date, units and definitions.
Check continuity
Look for missing or duplicated months, inconsistent formats and any revisions to previously published values.
Plot the trend
Use a line chart so the time order is visually clear.
Compare movements
Identify larger changes, stable periods and the difference between monthly and annual inflation measures.
Interpret the release
Use category-level evidence when explaining movements; do not infer a cause from the headline CPI series alone.
KNBS also publishes downloadable statistical tables covering areas such as agriculture, labour, finance, public finance, international trade, energy and education, so the same method can be extended to other Kenyan indicators.[7]
Finish with
A source table, cleaned dataset, trend chart, calculations, written interpretation, limitations and source notes.
Project 5: Perform a data-quality audit
Intermediate · Python or Excel — Find data problems before they contaminate the analysis.
Create a synthetic school-attendance dataset containing deliberate problems: duplicated rows, blank statuses, inconsistent class labels and invalid dates. Current early-career analytical work can explicitly require careful checking and reliable data handling before results are shared.[1][2]
| learner_id | class | date | status |
|---|---|---|---|
| L001 | F2A | 2026-02-01 | Present |
| L002 | form 2A | 2026-02-01 | P |
| L002 | form 2A | 2026-02-01 | P |
| L003 | F2A | invalid | Absent |
Synthetic training data only. No real learner records are used.
pandas provides documented methods for identifying missing data and working with tabular information.[8] Investigate the meaning of a problem before automatically dropping the affected rows.
| Test | Records affected | Action |
|---|---|---|
| Exact duplicates | 17 | Investigated |
| Missing attendance status | 31 | Flagged |
| Invalid dates | 4 | Sent for correction |
| Inconsistent class labels | 52 | Standardised after confirming meaning |
Protect personal data
For a public portfolio, use synthetic data, open public data, authorised data or appropriately anonymised data where lawful and suitable. Do not publish real student, employee, patient or customer records merely to demonstrate a project. Kenya’s data-protection principles include lawful and transparent processing, purpose limitation, data minimisation, accuracy and appropriate safeguards.[9]
Finish with
The raw synthetic dataset, audit script or workbook, a quality summary, cleaning log, cleaned version and a short explanation of your decisions.
Project 6: Analyse Kenya’s public-procurement data
Intermediate · Excel, SQL or Power BI — Work responsibly with a large public administrative dataset.
Kenya’s Public Procurement Information Portal publishes public procurement information, including contract records.[10] Kenyan PPRA data is also available through the Open Contracting Data Registry in reusable formats including CSV, Excel and JSON.[11]
Ask: How are published contract awards distributed by procuring entity, procurement method or time period?
Choose a manageable scope
Use one financial year, category, county or selected group of procuring entities rather than every record at once.
Inspect the export
Confirm what fields such as procuring entity, supplier, value, method and date mean in the downloaded data.
Audit quality
Check missing names, inconsistent text, extreme values and duplicated-looking records before aggregation.
Describe the pattern
Compare award counts by month, procurement-method frequency, entity totals and the distribution of award values.
Investigate anomalies
Treat unusually large values or concentrations as prompts for verification, not as proof of misconduct.
The Open Contracting registry exposes data-quality information for the Kenyan publication, which makes quality checking part of the project rather than something to hide.[11]
Evidence is not an allegation
A defensible statement is: “Supplier X had the highest aggregate published award value in the selected dataset and period.” A high value by itself does not prove fraud or corruption.
Finish with
The source and period, data-quality notes, cleaned data or queries, descriptive analysis, charts, exact definitions, limitations and no unsupported allegations.
Project 7: Build an end-to-end Data Analyst capstone
Intermediate · Multiple tools — Bring the entire analytical process together in one case study.
Imagine a fictional Kenyan retailer provides sales transactions, branch information, product costs, sales targets and customer records. Management asks: Why did gross profit change between Quarter 2 and Quarter 3, and where should the company investigate first?
Define the comparison
Specify Q3 versus Q2 and identify the branches, categories and measures to examine.
Audit the records
Check unique IDs, missing keys, duplicates, dates, unexpected negative values and inconsistent categories.
Build the analytical dataset
Use SQL to join only the tables you need, keeping the grain clear.
Separate the measures
Calculate Revenue, Cost, Gross Profit and Gross Margin. Revenue can rise while margin deteriorates.
Investigate the drivers
Test whether volume, selling price, cost, one branch, one category or a quality problem explains the movement.
Build the report
Show the smallest set of visuals needed to answer the management question.
Write the conclusion
State the evidence, identify where to investigate next and avoid turning an association into a proven cause.
State what remains unknown
Name missing information—such as supplier contracts or promotional expenditure—that could change the explanation.
Example conclusion
“Gross profit declined between Quarter 2 and Quarter 3 despite relatively stable revenue. Most deterioration occurred in Category B, where recorded unit costs increased while selling prices changed comparatively little. Branch C contributed a substantial share of the decline and should be investigated first.”
Which project should you build first?
| Your current skill | Start here |
|---|---|
| New to Data Analysis | Project 1 — Excel |
| Know Excel and basic SQL | Project 2 — SQL |
| Learning Power BI | Project 3 — Power BI |
| Want real Kenyan public data | Project 4 — Inflation |
| Know basic pandas | Project 5 — Data quality |
| Comfortable with larger public datasets | Project 6 — Procurement |
| Comfortable with several tools | Project 7 — Capstone |
The sequence is not compulsory. Choose the project that adds evidence you cannot yet show.
Before publishing a project: use this 10-point audit
Give yourself one point for each “yes”. The score is not a certification; it simply exposes what is missing.
Clear question?
Can you state the problem in one sentence?
Source identified?
Can another person tell where the data came from and whether it is public, authorised or synthetic?
Grain understood?
Do you know exactly what one row represents?
Quality inspected?
Did you check missing values, duplicates, categories, dates and unusual records?
Calculations explainable?
Can you explain every important formula, measure or query?
Totals checked?
Did you independently verify important numbers instead of trusting one tool blindly?
Visuals purposeful?
Does each important chart answer a question?
Findings readable?
Could a non-technical reader understand the main result?
Evidence separated from recommendation?
Did you distinguish what the data shows from what someone should do next?
Limitations stated?
Did you explain what the dataset cannot prove?
Can you defend the project in an interview?
Close the tutorial and answer these questions without reading your notes:
- What question were you trying to answer?
- Why did that question matter?
- What does one row represent?
- Where did the data come from?
- What was wrong with the raw data?
- What did you change?
- Why did you choose that method?
- How did you check the result?
- What was the most important finding?
- What could change your conclusion?
If you cannot answer these yet
Keep working on the project before adding another one. A portfolio is stronger when the work is yours and you can defend the reasoning.
Where can you find credible Kenyan datasets?
Kenya National Bureau of Statistics (KNBS): official economic, demographic and social statistics, including downloadable statistical tables.[6][7]
Kenya National Data Archive (KeNADA): an official catalogue of Kenyan survey and microdata resources; check each study’s access conditions before use.[12]
Public Procurement Information Portal: published public-procurement information suitable for carefully scoped contract and procurement-method analysis.[10]
World Bank Open Data: downloadable Kenyan indicators across economic, social and development subjects.[13]
Energy and Petroleum Regulatory Authority (EPRA): official energy and petroleum statistics that can support energy-trend projects.[14]
Synthetic data is also valid for learning
You can create fictional sales, attendance, customer-order or operations data when a real dataset would expose confidential information. Label it clearly as synthetic and never present invented values as official Kenyan statistics.
Common mistakes that weaken beginner projects
A beautiful report with no question
Start with the problem, not the colours.
A copied tutorial
Use tutorials to learn, then make independent choices in your own version.
A result you cannot reproduce
You should be able to explain how every important number was produced.
Too many tools
Do not add Python or another tool merely because it sounds advanced.
Machine learning for appearance
A clear SQL analysis can be more relevant to a Data Analyst role than an unexplained model.
Private information published publicly
Use public, authorised, anonymised or synthetic information instead.
Unsupported accusations
An anomaly deserves investigation; it is not automatically evidence of wrongdoing.
Frequently asked questions
What is the easiest Data Analyst project for a beginner?
Start with an Excel sales-analysis project. It lets you practise cleaning, formulas, PivotTables, charting and interpretation without first learning programming.
What SQL project should a beginner build?
A customer-and-orders database is useful because it teaches filtering, grouping and joins using a structure similar to many business databases.
What Power BI project is good for beginners?
Build a branch or sales-performance report using separate Sales, Product, Branch and Date tables. This lets you practise both reporting and basic data modelling.
Do I need Python before building Data Analyst projects?
No. Excel, SQL and Power BI can support strong beginner projects. Python becomes especially useful for repeatable cleaning, automation and programmable analysis.
Do I need a portfolio to become a Data Analyst?
Not every employer requires a public portfolio. Projects are useful because they give an employer concrete evidence to inspect when you have limited professional experience. A current Nairobi junior Data Analyst internship explicitly recognises academic projects and personal portfolio work as forms of hands-on experience.[2]
How many Data Analyst projects do I need?
There is no official number. Build projects that demonstrate different abilities. Three distinct projects you can explain and defend can be more useful than ten nearly identical dashboards.
Can I use fictional data?
Yes. Synthetic data is useful for practising business problems without exposing confidential records. Label it clearly as synthetic.
Should every project have a dashboard?
No. A SQL case study, data-quality audit or Python notebook can be a strong project without an interactive dashboard.
Can I use AI to help with my projects?
Yes, as an assistant. Use it to explain an error, generate synthetic practice data, suggest test cases or clarify syntax, but verify the output and do not publish work you cannot explain.
The final test: can someone see how you think?
The dashboard is not the whole project. The code is not the whole project either.
Question → trustworthy data → method → check → evidence → next step → limitation
A strong beginner project makes that reasoning visible. Build that habit from your first project and every later project becomes easier to defend.
For the wider career path, employer expectations and entry-level roles, continue to NIT’s Data Analyst Career in Kenya guide.
Course information below reflects NIT’s current Data Science & Analytics programme page.[15]
Data Science & Analytics at NIT
Learn the tools, then practise them on complete analytical problems
Newton Institute of Technology offers practical Data Science and Analytics training covering Excel, statistics, SQL, Python, data cleaning, visualisation and introductory machine learning.
Institutional training6 months
Industrial attachmentCompulsory 3 months
Study modePhysical + online
School feeKSh 60,000
AdmissionsContinuously open
Newton Institute of Technology · Migori Town, Kenya
Sources and references
- Moko Home + Living — Careers: Junior Data Analyst Internship. Current Kenya-based role context: turning questions from sales, operations and finance into numbers, reports and recommendations.
- MyJobMag — Junior Data Analyst Internship, Moko Home + Living (27 October 2026). Current evidence that academic projects or a personal portfolio can count as hands-on analytical experience, with SQL, pandas/Python and visualisation among requested skills.
- PostgreSQL Documentation — SQL Tutorial. Primary technical documentation for SQL querying; related PostgreSQL tutorial pages cover joins and aggregate functions.
- Microsoft Learn — Dashboards for Power BI business users. Primary technical source for the distinction between a Power BI report and a single-page dashboard in the Power BI service.
- Microsoft Learn — Understand star schema and the importance for Power BI. Primary technical source for fact tables, dimension tables, relationships and consistent grain.
- Kenya National Bureau of Statistics — Consumer Price Indices and Inflation Rates, September 2026. Official source for September 2026 CPI and inflation information.
- Kenya National Bureau of Statistics — 2025 Statistical Abstract. Official source for downloadable Kenyan statistical tables across multiple sectors.
- pandas Documentation — Working with missing data. Primary technical documentation for identifying and handling missing tabular data.
- Office of the Data Protection Commissioner — Rights of a Data Subject / Data Protection Principles. Official Kenyan guidance supporting lawful, purpose-limited, accurate and appropriately protected handling of personal data.
- Kenya Public Procurement Information Portal — Contracts. Official portal for published Kenyan public-procurement contract information.
- Open Contracting Data Registry — Kenya Public Procurement Regulatory Authority. Registry page providing Kenyan PPRA procurement data in reusable formats and data-quality information.
- Kenya National Data Archive (KeNADA). Official KNBS catalogue for Kenyan survey and microdata resources.
- World Bank Open Data — Kenya. Downloadable Kenyan economic, social and development indicators.
- Energy and Petroleum Regulatory Authority — Statistics. Official source for Kenyan energy and petroleum statistical reports.
- Newton Institute of Technology — Data Science and Analytics Course in Kenya. Current NIT programme information: six months of training, compulsory three-month attachment, physical and online study, KSh 60,000 fee and continuous admission.