Home / About / Data science

Capability

Data science

We analyse data so organisations can measure what is happening, test what works, and report with methods that can be checked.

Work covers descriptive statistics, inferential tests, survey analysis, forecasting, machine learning where it is justified, and graphs that a non-specialist can use in a meeting.

What we do

Analysis that can be repeated

Descriptive and inferential statistics

Means, rates, confidence intervals, hypothesis tests, and regression for study and operations data.

Survey and epidemiology analysis

Weighted surveys, case–control and cohort summaries, survival analysis, and outbreak or programme indicators.

Modelling and forecasting

Time series, classification, and prediction when there is enough data and a clear decision the model must support.

Visualisation and reporting

Publication-ready figures, dashboards, and tables exported to Word, PDF, Excel, or a web view.

Reproducible workflows

Scripts, notebooks, and documented code so the same result can be produced again from the same files.

Training and review

Walk-throughs of methods with your team, and review of existing analysis for errors or unclear assumptions.

Tools we use

Statistical software

RPrimary statistics language — tidyverse, ggplot2, survey, survival, epiR, lme4
Pythonpandas, NumPy, statsmodels, scikit-learn, Jupyter for analysis and pipelines
StataDo-files for survey, panel, and health datasets used by many research partners
SPSSSyntax-based analysis where a partner already works in SPSS
ExcelChecks, pivot tables, and handover files — not a substitute for scripted analysis
Jupyter / R MarkdownNotebooks and knitted reports that keep code, numbers, and narrative together
Epi InfoField epidemiology summaries when a study is already set up in Epi Info
GitVersion control for analysis scripts so changes are traceable

Need this capability on a live project?

Tell us the dataset, the field setting, and the decision the work must support. We will propose a concrete stack and a delivery path.

Contact SAWE