Data Analyst based in Nairobi, Kenya — turning messy, complex data into clear decisions.
🌐 Portfolio: collinsanyona.github.io — projects, findings, résumé, and contact in one place.
I work across the full analytics lifecycle: data ingestion, cleaning, pipeline automation, statistical modelling, and dashboard development. My background spans agri-tech, research, fintech, and NGO/social impact environments, and I care about building analysis that actually gets used.
| Category | Stack |
|---|---|
| Languages | Python · R · SQL (PostgreSQL, MySQL) |
| Visualisation | Power BI · Tableau · Excel Dashboards |
| Analytics | EDA · Statistical Analysis · NLP / Topic Modelling |
| Data Engineering | ETL Pipelines · Data Cleaning · Data Quality |
| Methods | M&E · Survey Design · Business Intelligence |
| Platforms | BigQuery · Git/GitHub · Google Sheets |
Growth-accounting analysis of six decades of African agricultural output: is growth coming from more inputs, or from farming better?
- Decomposed output growth into input vs productivity contributions across 200+ countries (USDA ERS data, 1961–2019)
- K-means typology of 52 African countries + trend forecast to 2035, with an honest limitations section
- Built in Python · pandas · scikit-learn | View Project →
Which African countries beat the child-mortality trend since 1990 — and why the SDG scoreboard was already written in 2015.
- Panel analysis of under-5 mortality across 54 African countries (UN IGME / World Bank data, 1990–2024): SDG 3.2 pace, beat-the-trend residuals, fixed-effects regression
- Honest ML result: a single-feature baseline ties gradient boosting — and the README says so
- Built in Python · pandas · statsmodels · scikit-learn | View Project →
Africa got children into school — did it get them through it? Enrollment vs completion across 42 Sub-Saharan countries.
- Gap metrics, gender-parity trajectories, and per-country trend analysis (1990–2024)
- Model-card forecast comparison where the naive baseline beats both linear trends and gradient boosting — knowing when simple wins
- Built in Python · pandas · scikit-learn | View Project →
Full-stack civic-tech platform making government budget data explorable for citizens, with AI-assisted insights and moderation.
- FastAPI + React platform: budget analytics dashboards, citizen feedback, forums, role-based admin
- Google Gemini integration for AI insights and content moderation; deployed live on Render
- Built in Python · FastAPI · React · PostgreSQL | View Project → · Live Demo
Founder project: a Flask platform giving Kenyan smallholder farmers an AI farming advisor, marketplace, and expert booking.
- OpenAI-powered chatbot for agronomic advice, plus product marketplace and hydroponics learning content
- End-to-end product: auth, bookings, forums, checkout — designed, built, and deployed solo
- Built in Python · Flask · OpenAI | View Project →
- Research Fellow, Equitech Futures — Built Python & R pipelines for large-scale research datasets; automated NLP workflows for topic modelling and insight extraction
- Founder, Shamba Salama — Designed end-to-end data collection and monitoring systems; built Power BI dashboards for agri-tech performance tracking
- Teaching Assistant, USIU-Africa — Taught data cleaning, EDA, and reproducible analysis workflows to undergraduate students
- BSc (Hons) Software Engineering — African Leadership University, 2026
- BSc Data Science & Analytics (in progress) — USIU-Africa
- Google Data Analytics Professional Certificate — Coursera, 2026
I'm currently open to Data Analyst, Business Intelligence, and Research Analyst roles — in Nairobi or remote.
"Data is only as valuable as the decisions it enables."