Quelle anzeigen

Machine learning Group 2021

Following the great interest in continuing the work of the HLG-MOS Machine Learning Project, UK Office of National Statistics (ONS) is launching the Machine Learning Group 2021. The new ML group will focus on developing and implementing ML for official statistics (see slides below for presentation and live survey conducted during the HLG-MOS ML Project webinar)

Table 1: Work packages investigated by the ML 2019/2020 project (hyperlinks) and potential themes to be explored by the ML 2021 project (text in red).

The Journey	Moving from idea to valid solution (demonstration)		Moving from valid solution to production (Operationalisation)		Ensuring production robustness (Maintenance)
	All WP1 pilot studies		Some WP1 pilot studies		Very few WP1 pilot studies
	Other applications of Machine Learning		Some other applications of Machine Learning		Very few other applications of Machine Learning
	WP3 Integration (Q5 & Q6)		WP3 Integration (Q5)
	Workstream 1: Support current studies towards production; welcome new studies in other processes (e.g. record linkage) and/or data sources (e.g. satellite data)
Supported by	Quality (accuracy, timeliness, efficiency, explainability and reproducibility)	Good Training Data	Skills/Competences	Computing Infrastructure	Interoperability / Business Process	Ethics and Legal	Security
	WP2 Quality		WP3 Integration (Q3 & Q4)
	Workstream 2: Experiment with practices and methods on some dimensions of QF4SA (WP2); Workstream 3: Review and improve the Framework	Workstream 4: How to get good training data, how to keep it up to date, when to relearn a model, what does 'good' mean, how to measure that?	Workstream 5: What skills? How to learn? Where to find them?	To be defined	To be defined	Workstream 6: Ethics handbook, regulations , etc.	To be defined
Facilitated by	Organisation			Sharing and Collaboration
	WP3 Integration (Q1 & Q2)			HLG-MOS Machine Learning Project
	Initiatives to accelerate the integration of machine learning solutions			ML Studies and Codes
	Workstream 7 : Create/maintain a network of data science unit leaders; Workstream 8: Beyond 2021: How can we better prepare for next 2-5 years? What technology and data sources can we expect? What skills will we need?			Learning and Training
				HLG-MOS ML Project webinar

How to join the Group ?

Do you have any ML topic you are interested in working together with peers? Do you have any issue (technical, strategic, organisational) you want to discuss with other NSOs? Contact ML2021 (ML2021 at ons dot gov dot uk), copying UNECE (choii at un dot org) if you want to join the Machine Learning 2021 Group!

If you are a member of the Global Network of Data Officers and Statisticians, you can follow us from the ML for Official Statistics group in the Network (see quick guide on how to join the Global Network).

Progress status

Update from ONS (December 15, 2020)

Resources

Machine learning competency in context of Big Data training and human resources (from UN Global Working Group on Big Data Task Team on Skills and it is relevant to the community)