Open-source, mostly offline-first tools I build for official statistics — for agricultural census methodology, survey data cleaning, small area estimation and guideline search. Source code, and a live demo where available, is linked under each tool.

Statistics tools

Tools for agricultural statistics and survey methodology

WCA 2030 Explorer

Offline progressive web app for querying the FAO WCA 2030 guidelines. Answers are verbatim extracted text with section and page citations — never generated. Useful for census methodology teams who need precise, citable references to the guidelines without an internet connection.

Vite Transformers.js (WASM) MiniSearch BM25 PWA

AC Metadata Review & Tables of Main Results (PIPELINE_AC_MR_TMR)

Desktop app for FAO WCA 2020 census methodology staff. Generates Metadata Reviews (15 sections) and Tables of Main Results (23 sub-tables) from national agricultural census documents using LLM APIs — reducing manual processing time for census assessments.

TypeScript LLM APIs

Data Cleaning Syntax App

Offline-first, open-source tool for survey statisticians. Turns codebooks and metadata into transparent data-cleaning, validation and imputation syntax for SPSS, Stata, R, and Python — making cleaning workflows reproducible and auditable without proprietary tools.

TypeScript Offline-first SPSS Stata R Python

SAE Syntax Generator

Offline-first progressive web app for statisticians. Upload a survey codebook, choose from 16 area-level and unit-level small area estimation methods, and download ready-to-run, commented R and Stata scripts — lowering the barrier to applying SAE in national statistical offices.

TypeScript PWA R Stata

Other projects

Data science portfolio

PIMA Diabetes Prediction

A predictive model that classifies whether a patient is likely to have diabetes from health attributes (PIMA Indians Diabetes Dataset). A data-science portfolio project demonstrating end-to-end machine learning — from exploratory analysis to model evaluation.

Python Machine learning

Source code

All repos on GitHub

All source code is publicly available under open-source licences. Contributions and feedback are welcome.

View GitHub profile