This fits you strongly. FinAccess is almost a perfect bridge between your data scientist side and your wanderer / representation-learning side.

It is not just a finance dataset. It is a dataset about how Kenyans are represented through access, usage, quality, welfare, mobile money, credit, savings, county, age, gender, rural/urban life, disability, shocks, and financial health. The official FinAccess site has the 2024 report plus questionnaire, manual, and anonymised weighted datasets in SPSS, Stata, and Excel formats. (finaccess.knbs.or.ke)

The report itself frames FinAccess around four dimensions: Access, Usage, Quality, and Impact/Welfare, and says the 2024 survey tracks financial inclusion nationally and at county level. That is already a representation system: it turns messy financial lives into measurable categories.

For your blog, I would make this a flagship series.

The angle that fits you

Not:

“Analysis of FinAccess 2024”

Better:

How Kenyans’ financial lives are represented in survey data

That connects to your interest in representation learning.

The survey asks: who is included, excluded, borrowing, saving, defaulting, betting, using mobile money, using SACCOs, facing fraud, financially healthy, investing in green assets, or coping with shocks. The 2024 headline findings include 84.8% formal financial access, 9.9% exclusion, daily mobile-money use at 52.6%, credit usage at 64.0%, savers declining to 68.1%, and financial health remaining low at 18.3%. (Kenya National Bureau of Statistics)

The real question for you is:

What hidden structure sits underneath these categories?

Possible blog series

1. FinAccess as a representation problem

Start with the idea that surveys do not merely “collect data.” They create a way of seeing the country.

Questions:

What does FinAccess choose to measure?
What does it leave out?
How does it represent inclusion, exclusion, health, risk, and welfare?

This fits your public participation / Taleb / ground-knowledge interests.

2. Mapping financial lives, not just financial access

Use clustering or dimension reduction.

Possible methods:

MCA for categorical variables
PCA for constructed scores
UMAP for visual exploration
clustering for financial-life typologies
decision trees for interpretable segmentation

Possible groups you may discover:

mobile-money-dependent
banked but financially stressed
SACCO/resilience group
excluded rural youth
digital-credit-exposed
high-access but low-health urban borrowers
informal-only network users

That is representation learning in a Kenyan social dataset.

3. The geography of financial inclusion

County-level maps are natural here. The report says Kiambu, Nairobi, Kirinyaga, Nyeri, Isiolo, and Mandera are among the most included counties, while Turkana, West Pokot, Elgeyo Marakwet, Trans-Nzoia, Migori, and Narok are among the most excluded. (Kenya National Bureau of Statistics)

You could ask:

Are excluded counties simply poor, or are there different types of exclusion?
Is mobile money flattening geography?
Do some counties have access without financial health?

4. Mobile money as infrastructure

This is very you. Mobile money is not just a product. It is a layer of national infrastructure.

The report says mobile money remains an equalizer across demographics, while also noting consumer protection problems such as system downtime, fraud, hidden charges, and money loss.

Possible post title:

Mobile Money as Kenya’s Hidden Public Infrastructure

5. Financial access versus financial health

This may be the strongest analytical question.

Access can rise while financial health remains weak. That is a serious development-policy issue. The report says formal access rose to 84.8%, but only 18.3% of Kenyans were financially healthy. (Kenya National Bureau of Statistics)

Possible title:

Kenya Is Financially Included. But Is It Financially Healthy?

That is clean, public-facing, and analytical.

How this can fit your homepage intro

This is exactly why your intro should mention representation. Something like:

I write to think through questions across data, health, technology,
institutions, markets, books, and ordinary life.

My formal home is statistics and data science. The wider thread is
representation: how reality becomes data, models, categories, maps,
institutions, and decisions. This site is a personal archive of technical
notes, learning notes, book thoughts, and small observations that help me
understand how things work.

FinAccess is a perfect example of that sentence. It turns Kenyan financial life into categories, indicators, maps, scores, and policy decisions.

Your first post could be:

Why I Want to Analyze FinAccess

With a thesis like:

FinAccess is not only a financial inclusion survey. It is a map of how
Kenyans interact with money, institutions, technology, risk, and the future.
For me, the interesting question is not only who is included, but what kind
of financial lives the data allows us to see.