About the meeting
The Machine Learning for Official Statistics Workshop 2023 aims to bring together experts in national and international statistical organisations to share developments in the field and discuss issues and challenges.
UNECE Machine Learning for Official Statistics Workshop 2023
05 - 07 June 2023
Geneva Switzerland
Timetable and information notice
Session 1: Machine Learning Applications
Session Organizers: Michael Reusens (Statistics Flanders, Belgium) and Joni Karanka (Office for National Statistics, UK)
| Classifying companies in France using machine learning - Thomas Faria and Tom Seimandi (Insee, France) | Abstract | Paper | Presentation |
| Using Webdata to derive the Economic Activity of Enterprises - Manveer Mangat (Statistics Austria) | Abstract | Paper | Presentation |
| Clothing Price Index using Web-Scraped Data - Laura Christen and Ahmet Aydin (Office for National Statistics, UK) | Abstract | Paper | Presentation |
| Imputation of occupation in the Occupational Register - Jens Malmros (Statistics Sweden) | Abstract | Paper | Presentation |
| Too good to be true? A case of machine learning in the validation process of the R&D statistics - Eva Charlotte Berner and Solveig Bjørkholt (Statistics Norway) | Abstract | Paper | Presentation |
| Geospatial Bayesian Methods for Hazard-Impact Modelling - Hamish Patten (University of Oxford) | Abstract | Paper | Presentation |
| Progression patterns in the Swiss social security system based on Machine Learning: methods for evaluating quality and model drift - Athanassia Chalimourda (Swiss Federal Statistics Office) | Abstract | Paper | Presentation |
| ML Poverty: Using Machine Learning to estimate poverty rates in Switzerland at the canton level - Yara Abu Awad (Swiss Federal Statistics Office) | Abstract | Paper | Presentation |
| Creating a modern business index: Machine learning record linkage at scale - Isabela Breton and Joanne Sheppard (Office for National Statistics, UK) | Abstract | Paper | Presentation |
| Time Series Outlier Detection using Metadata and Data Machine Learning in Statistical Production - Olivier Sirello (BIS) | Abstract | Paper | Presentation |
| Timeliness and Accuracy with Machine Learning Algorithms: Early Estimates of the Industrial Turnover Index - David Salgado (Statistics Spain) | Abstract | Paper | Presentation |
| Nowcasting TiVA indicators: improving timeliness of trade data - Polina Knutsson (OECD) | Abstract | Paper | Presentation |
Session 2: Quality Aspects of Machine Learning in Official Statistics
Session Organizers: Florian Dumpert (Federal Statistical Office of Germany) and Ralf Becker (UN Statistical Division)
| Quality Framework for Statistical Algorithms - InKyung Choi (UNECE) | - | Paper | Presentation |
| A Quality Concept for the Use of Machine Learning in Official Statistics - Florian Dumpert (Federal Statistical Office of Germany) | Abstract | Paper | Presentation |
| Exploring quality dimensions in trustworthy Machine Learning in the context of official statistics: model explainability, accuracy and uncertainty - Saeid Molladavoudi (Statistics Canada) | Abstract | - | Presentation |
| Understanding model quality in the context of trustworthiness and value - Emily Barrington (Office for Statistics Regulation, UK) | Abstract | Paper | Presentation |
| Lessons learned when applying Machine Learning in Official Statistics: Why it helps to be a survey statistician and a data scientist! - Piet Daas (Statistics Netherlands) | Abstract | Paper | Presentation |
| Changing Data Sources in the Age of Data Science for Official Statistics - Cedric De Boom (Statistics Flanders, Belgium) | Abstract | Paper | Presentation |
Session 3: Toward system-wide transformation of statistical production
Session Organizers: Riitta Piela (Statistics Finland) and Dominika Nowak (Statisitcs Poland)
| Keynote presentation: How to Manage Machine Learning as a Process of Continuous Improvement in the Context of Official Statistics - Prof. Diego Kuonen (Statoo Consulting & GSEM, University of Geneva, Switzerland) | - | - | Presentation |
| Facilitators and Blockers of ML Adoption in Official Statistics - Joni Karanka (ONS, UK) | Abstract | Paper | Presentation |
| A Machine Learning Capability Uplift Strategy - Claire Clarke (Australian Bureau of Statistics) | Abstract | - | Presentation |
| ML training : Who? What? How? and… What for? - Christophe Bontemps (UN Statistical Institute for Asia and the Pacific, ESCAP) | Abstract | - | Presentation |
| Balsam: A Collaborative Platform to Support ML and ML-Ops initiatives - Jakob Engdahl (Statistics Sweden) | Abstract | Paper | Presentation |
| An open source data science platform to foster innovative and production-ready machine learning systems - Romain Avouac (Insee, France) | Abstract | Paper | Presentation |
| Hands-on Lab: An introduction to MLOps with Mlflow - Tom Seimandi, Romain Avouac and Thomas Faria (Insee, France) | Abstract | - | Presentation |