What can Machine Learning offer as a service to accelerate National Statistical Office Transformation?
Developments in machine learning (ML) are pushing the boundary of tasks that were considered tasks for humans - machines can now understand text, recognise objects in images, and even write news articles, just like humans. The technology has great potential to transform the way statistical offices work. It can increase the efficiency of production by automating the manual processes. It can also improve the relevance of statistics by making it possible to exploit new data types. The use of ML is particularly important for big data as processing such a large amount of data using traditional methods is often prohibitively costly and time-consuming.
This session provides an introduction to machine learning, presents the practical applications of machine learning in working areas of statistical offices, and discusses challenges that statistical offices face when using machine learning.
Agenda Download ML Group Report
Presentation
Machine Learning for Official Statistics Presentation
This presentation provides a brief introduction on machine learning for official statistics.
What is machine learning? How is it different from traditional statistical methods? Why has it become so important these days? Can it be really used for official statistics, or is it just a hype? How can organisations start exploring machine learning? What support can international initiatives such as the ONS-UNECE ML Group 2022 offer to help NSOs seeking to upgrade their Data Science capabilities?
Presentation
Machine Learning Use Cases from National Statistics Offices
Like any technology, machine learning is one possible means to an end, and as such, it should not be considered or adopted simply for what it is, but for what it can do to better address the business needs in the statistical offices. This presentation demonstrates three ML use cases that are increasingly popular in the National Statistical Offices (NSOs) – edit and imputation, coding and classification, and imagery analysis.
- Satellite Imagery to map poverty in Indonesia Presentation
- Social Surveys at UK ONS Presentation
- Text Classification For Real Estate Data at Statistics Poland Presentation
Discussion
Challenges in Using Machine Learning in National Statistics Offices
Integrating ML in production comes with many challenges and setbacks. Unfortunately, many machine learning solutions, even after successful pilot studies, end up being left on the shelf. Why is it so difficult to push ML into production? This discussion examines common challenges that NSOs encounter when using ML such as capacity, IT, culture and organisational structure.
Training session - Novice Level
Machine Learning Fundamentals Presentation
What does Machine Learning do, what are the basic principles, demonstration
Open to all. Some coding experience preferred.
Lunch
Training session - Intermediate + Advanced Level
Coffee and Coding I - Machine Learning Applications Presentation
A chance to get hands on experience of technical processes of how ML applications are developed and used.
Training session – Advanced Level
Coffee and Coding II – Coffee and Coding Applications Deep Dive Presentation
In this session participants will be given a deep dive into the application of ML algorithms in practice together with python code examples.