Data Science

NumPy: Whole Columns at Once

Machine learning is maths on thousands of numbers at once. NumPy arrays let you do it in one line — no loops — and they're what every ML library runs on underneath.

pandas: Tables You Can Question

pandas is the spreadsheet of data science. Load 80 farms, select, filter, group and clean them — the everyday moves before any model is trained.

Visualising Data

A chart shows in a second what a table hides. Learn to pick the right chart for the question and draw it with matplotlib — scatter, bar, line and histogram.

Exploratory Data Analysis

Before any model: get to know your data. Summaries, distributions, outliers and correlations — and the discipline to say what the data does and doesn't show.

Framing a Data Question

A county health officer asks: "Where is malaria worst?" Before any code, turn that into a question data can answer — and discover how the choice of denominator changes the answer.

Cleaning Real-World Data

Clinic records typed by busy staff are full of double entries, spelling variants, "999 = unknown" codes and unit slips. Clean them with rules you can defend — and count what every rule changed.

Joining & Reshaping

Answers usually need more than one table. Join visits to clinics and populations without losing or doubling rows, then reshape the result between long and wide to answer month-by-month questions.

Dates, Text & Categories

Three date formats in one column, a pandas shortcut that silently gets hundreds of them wrong, and ages that need grouping. Parse, extract and categorise with care — then find the malaria season.

Clinic Records Clean-up

A regional health office gets messy monthly exports from 18 clinics. Build the clean-up pipeline, a data-quality report to send back to the clinics, and a malaria summary they can trust.

Distributions & Outliers

Before any average, look at the whole distribution. Explore 600 incomes: their skew, why mean and median disagree, which values are outliers — and how urban and rural applicants compare.

Relationships

Two variables moving together can mean a lot or nothing. Plot before you calculate, measure correlation properly, and meet the paradox where every group says one thing and the total says the opposite.

Designing Clear Charts

A correct chart isn't the same as a clear one. Turn a default bar chart into one a busy decision-maker understands in five seconds, keep the axes honest, and compare six counties at a glance.

Statistics That Matter

Mean, median, standard deviation, the normal distribution, and why a sample never tells you the exact truth — the statistics every model result leans on.

Probability & Bayes

Why a positive test can still be unlikely, how spam filters think, and how to reason about uncertainty with simulation and Bayes' rule.

Samples & Populations

Surveys ask a few thousand people and describe millions. See why that works — sampling error shrinks predictably — and why it fails when the sample is biased, however big it gets.

Confidence Intervals & the Bootstrap

An estimate without uncertainty is half an answer. Build confidence intervals for a proportion, check what "95%" really promises, and use the bootstrap to put error bars on anything — even a median.

Hypothesis Tests

A new maize variety yields 2 more bags per acre on trial plots. Real, or luck? Build a permutation test from scratch, check it against the t-test, and learn why testing everything guarantees false discoveries.

Did the Programme Work?

A county gave 20 struggling schools extra support, and their exam scores jumped. Before the county scales it up, work out how much of that jump the programme caused — and how sure anyone can be.

A/B Tests

A mobile-money provider wants to know whether an SMS reminder makes customers save. Randomise, check the groups are balanced, estimate the lift with a confidence interval — and work out how many users you'd need.

Confounding

Students who get private tuition score 9 points higher. Does tuition cause that? Find the confounder, compare like with like, and adjust with regression — and see how much of the gap survives.

Natural Experiments

Machakos launched a community health programme and child immunisation rose. But it was rising before, and in neighbouring Makueni too. Use difference-in-differences to separate the programme from the trend.

The Savings Reminder Trial

A mobile-money provider ran a randomised trial of two SMS reminders. Analyse it properly — balance, lifts with intervals, what the trial can and can't tell apart — and advise on the rollout.

Excel & Spreadsheets

Most organisations still run on spreadsheets. Read a real-looking office workbook — title rows, totals, lost leading zeros and all — tidy it with pandas, and hand back a clean Excel report your colleagues can open.

SQL for Analysts

Real data lives in databases, and SQL is how you ask for it. Build a database in your browser, then filter, group, join and clean with SQL — and hand the answers to pandas.

Survey Data & Weights

Household surveys deliberately over-sample small areas, so a plain average is wrong. Use design weights to turn a sample of 760 households back into an estimate for all 5,000 — and for totals, not just shares.

Maps & Location Data

Where something happens is often the story. Draw Kenya's 47 counties from real boundary data, map population density honestly, and measure distances between places on a round Earth.

County Services Survey Report

A county planning office commissioned a household survey. Produce the official estimates — weighted, with uncertainty — and deliver them as an Excel workbook and a map they can put straight into their plan.

Storytelling with Data

Kenya went from 15% to 76% electricity access in two decades — the fastest gain among 20 African countries. Turn real World Bank data into a message a busy reader gets in one glance: the right numbers, in context, on an annotated chart.

Reports & Dashboards

Some questions come back every month. Design a dashboard for a county health director, build it in Python, and publish it as a shareable report — and learn where Power BI, Tableau, Looker Studio and notebooks fit.

Open-Data Investigation

Your final project uses real open data. An energy-access charity asks: which African countries made the fastest progress on electricity since 2000 — and did rural areas keep up? Investigate the World Bank's data end to end and deliver a brief, a chart and a workbook.