Zum Ende der Metadaten springen

Machine Learning

Progress

(The portions in blue below are links that should lead you to documents providing more details on the Machine Learning project Home page)

There are now 29 participants from 14 organisations and 12 countries. A profile of the participants and some machine learning projects conducted in their organisation was produced and shared. The participants reported 39 machine learning projects on a wide range of data sources, processes, objectives and level of maturity. Another 5 persons were added to a distribution. They will also contribute to the project.
Given the interest of the majority of members to work on pilot studies, the planning focused on WP1
A planning meeting was held on April 4, followed a by progress meeting on April 16.
Leads for the three work packages were identified.
The objectives of the project were set. The ultimate objective is to advance the research, development and application of machine learning techniques to add value to the production of official statistics.
It was agreed to conduct pilot studies (WP1) on four themes:
- Classification and coding on more traditional data (survey, census, administrative/register); led by UK - ONS
- Edit and imputation on more traditional data; lead to be identified
- Imagery; led by Mexico - INEGI
- Web and sentiment data; led by Poland - Statistics Poland
A lead organisation was identified for three of the four pilot study themes
Each organisation was asked to proposed at least one machine learning project to be considered for one of the pilot study themes. A form was developed to capture baseline information on these projects.
Inference with Big Data, namely sensor data, and related challenges was reported by several organisations; it will be addressed in WP2 on data quality
The ONS have booked venues in London and Newport to hold a sprint meeting from May 13-15.

Next Steps

Obtain baseline information on each machine learning project proposed by the organisations to finalize the specific objectives and tasks of each WP1 pilot study
Set the detailed objectives and tasks of the WP2 on data quality
Conduct a sprint hosted by the ONS May 13-15. So far, it would be attended by 11 participants from 8 organisations in 7 countries. Facilities for others to connect and participate will be arranged. The ONS has submitted a paper to provide an overview of the ML project and the outcome of the sprint to the ModernStats workshop.

Risks and Issues

Issue	Mitigation

News from the Groups
Blue-skies Thinking	Identifying Topics/Opportunities	IN PROGRESS
Blue-skies Thinking
Developing Organisational Capability	Skills and Capability Framework	IN PROGRESS	We are planning to start with identifying complementary skills. The top level of the GAMSO framework could be used as a starting point for the work.
	Promotion Forum	IN PROGRESS	We need to update public wiki site of our Group regularly, and looking forward to getting proposals on fresh blog posts. We also need feedback on: what will draw you to this wiki site? and what you find the most useful? We can also consider to share national news on the blog and to add Discussion Forum, if useful. We will advertise public wiki page during the various UNECE Workshops, All Group members are also encouraged to share link to the wiki page with their colleagues. We are planning to prepare some handout about the work/outputs of our group for the CES session this June, maybe presentation at the CES with promotional video (it should be checked). In near future our promotional material should be interactive, short, sharp and action focused. We need to focus on the benefits we add to the NSO's, and demonstrate the value of the outputs to the higher degree.
	Setting vision in NSOs	IN PROGRESS	We are planning to start work with the short questionnaire on how to define the best vision for NSO's. We need more people interested to contribute to the work on Setting Vision in NSO's.
	Other	We are under the process of forming Organising Committee of the workshop on Culture Evolution. Andrew Mann from ABS agreed to be a chair of the Workshop. All members of this Group are asked to volunteer to be co-organisers of one of the proposed sessions of the Workshop, and when OC is formed, we will schedule separate, more frequent, calls to prepare this Workshop. It is proposed to have 2 co-organisers for every session.
Supporting Standards	Linking GSBPM and GSIM	IN PROGRESS	The task team, jointly chaired by Istat and Statistics Canada, had started its work. A template has been agreed to map GSIM information objects to GSBPM sub-processes as well as to provide concrete examples. Two GSBPM sub-processes have been mapped by Stats Canada and Istat. The Group has volountereed to produce the mapping for a few sub-processes to be discussed at next meeting. The task team is meeting every three weeks.
	Core Ontology	IN PROGRESS
	Alignment GSBPM and GAMSO	IN PROGRESS
	Metadata Glossary	IN PROGRESS
	Other
Sharing Tools	Digitizing/editing CSPA document	IN PROGRESS
	Adding Services to Catalogue	IN PROGRESS
	Communicationg restated CSPA	NOT STARTED
	Other