Independent practice · Aydin
Thoughtful analysis.
Evidence you can work with.
I help researchers and teams choose appropriate statistical methods, understand their results, and make practical use of their data. My background combines data science and strategy with doctoral research into longitudinal statistical methods at the University of Sydney.
Start with the question, examine the data, and make the assumptions visible. The aim is an analysis that can be understood, checked, and used.
A selection of projects from previous roles across technology, government, energy, retail and media.
These projects were completed in previous employed roles, rather than as Pebble Rocket engagements.
Developed analytical approaches to government compliance questions, combining statistical modelling, machine learning and exploratory analysis. Built tools within client teams to support investigations and translated findings into recommendations for targeted intervention.
Led a team analysing consumer switching in the energy sector. The work examined who switched providers and how switching affected energy bills, providing evidence for policy development.
Rebuilt core analytical processes behind a large media dataset, including generalised linear models for estimating consumption factors. Worked with client stakeholders and software developers to introduce changes without disrupting related analytical work.
Developed statistical methods for business performance metrics and contributed to a reporting tool for a large Australian retailer. The work also included clustering analysis and analysis of business and product performance.
Established a data science function supporting company-wide engineering priorities. Defined measurement needs and oversaw projects investigating engineering effectiveness, identifying opportunities for improvement and supporting business performance management.
Ten years of experience in data science, with R and SQL as part of my practice. My work also spans strategy and team-building.
I am a doctoral researcher at the University of Sydney, working on longitudinal statistical methods in a healthcare research setting.
Aydin Sareff-Hibbert and Gillian Z. Heller
Preprint · First posted in 2024; revised November 2025.
A comparison of statistical approaches for analysing correlated outcomes.
Software for longitudinal statistical modelling, with Aydin as author and maintainer. The accompanying paper is a work in progress.
For statistical consulting enquiries, email pebblerocket@proton.me. Include a short description of your question, the data involved and your timeframe. Please do not send confidential data in an initial email.