01 / PREDICTIVE MODELING
House Price Prediction
Estimate California median house values while comparing a transparent baseline with a nonlinear ensemble model.
- 20,640
- records
- 8
- predictors
- RMSE
- primary metric
Open to data analyst and AI data roles
From raw rows to a decision-ready story.
I build analytical models, dashboards, and reporting workflows that make complex data useful. This portfolio follows three end-to-end projects from question to method to decision.
- Portfolio
- 3 case studies
- Toolkit
- Python / SQL / BI
- Based in
- Lagos / Remote
+ signalpatterns isolated
01decision framed
PYTHON SQL EXCEL POWER BI MACHINE LEARNING DATA STORYTELLING
Selected work
Three questions. Three analytical lenses.
Each case study is designed to show business framing, clean methodology, and honest interpretation.
02 / PUBLIC HEALTH BI
COVID-19 Trends Dashboard
Turn country-level reporting into weekly trends while making reporting gaps and revisions visible to the reader.
- 2020+
- coverage
- Weekly
- reporting view
- 4-week
- trend window
03 / FINANCIAL ANALYTICS
Stock Price Analysis
Profile returns, moving averages, volatility, and drawdowns from daily market data without making price forecasts.
- Daily
- OHLCV
- 20 / 50
- day averages
- Risk
- not advice
Case study 01
What makes a house-price model useful rather than merely accurate?
A reproducible comparison between linear regression and random forest, with leakage-safe validation and error analysis by value band.
Business question
Which property and neighborhood signals explain median value, and where does the model make costly errors?
Method
Train/test split, median baseline, two candidate models, RMSE / MAE / R², residual checks, and feature importance.
Decision output
A model card that separates performance from limitations and identifies the segments that need human review.
MODEL DIAGNOSTICActual vs. predicted
Illustrative interfacePortfolio interface preview. The downloadable analysis calculates final metrics from the source dataset.
Case study 02
How do you show a pandemic trend without hiding the reporting problem?
A weekly dashboard that prioritizes comparability, revisions, and data completeness alongside cases and deaths.
Business question
Where are reported trends changing, and how confident should a reader be in comparisons across countries?
Method
Schema checks, country filters, weekly aggregation, four-week averages, per-capita options, and correction flags.
Decision output
A dashboard with trend, comparison, and data-quality views plus plain-language caveats.
WEEKLY TRENDCases with smoothing
Reporting-awareWHO notes that reporting frequency, definitions, and retrospective corrections differ across countries.
Case study 03
What does price history say about behavior and risk - not tomorrow?
A disciplined market-data workflow focused on returns, trend regimes, rolling volatility, and drawdown.
Business question
How has the security behaved across trend and stress periods, and what risk is visible in the historical series?
Method
API ingestion, OHLCV validation, daily returns, 20/50-day averages, annualized volatility, and max drawdown.
Decision output
A concise risk profile with reproducible indicators and no unsupported investment recommendation.
NORMALIZED VIEWPrice + moving averages
For analytical demonstration only. Historical patterns are not investment advice or a forecast.
Working method
Analysis that can survive a second look.
- 01
Frame
Translate the request into a decision, audience, metric, and known constraint.
- 02
Validate
Profile the source, document definitions, test assumptions, and surface missingness.
- 03
Analyze
Choose the simplest credible method, benchmark it, and inspect where it fails.
- 04
Communicate
Package the result in a dashboard or brief with caveats a reader can actually use.
About the analyst
Data rigor, business context, clear writing.
I am a data analyst with a Data Science Diploma and experience in Python, SQL, Excel, Power BI, machine-learning evaluation, data labeling, and campaign analytics.
My work combines analytical detail with practical communication: dashboards, written briefs, documented workflows, and recommendations for non-technical stakeholders.
PythonpandasSQLPostgreSQLExcelPower QueryPower BIDAXStatisticsML evaluation
Let us work together