Machine Learning

Progress

A virtual sprint was held from March 31 to April 16. Twelve sessions were organized and attended by 60 participants from 24 organisations throughout the period. We thank all organizers, presenters, participants and the UNECE Secretariat for a very successful exchange and progress in this challenging situation.
The project welcomed its Champion, Anders Holmberg from the HLG-MOS Executive Board. The virtual sprint generated additional interest with 9 collaborators and followers connecting to the project from Australia, Montenegro, UK, Netherlands, Sweden and the IMF. The project now has 39 participants and 51 collaborators/followers.
Following the sprint, it was decided that the project was ready to start "going public" at the Machine Learning for Official Statistics page. The page currently introduces the project, shares most of the presentations given at the virtual sprint and announces plans for upcoming outputs.
There has been significant progress made on creating a hands-on ML application that will serve as a learning tool. It is based on the coding of web-scraped product description made available by Statistics Poland.
Monthly update meetings regularly attract 30 participants. At the March meeting, Taeke Gjaltema gave an update on the UNECE's position regarding the ML project. In spite of the COVID-19 situation, important staff shortages at the UNECE Secretariat and the fact the that most project participants are working from home, it continues to fully support the ML project. This message was much appreciated by the ML group.
The session on 'The Integration of Machine Learning into Official Statistics' for BigSurv20, Continuing to Explore New Statistical Frontiers at the Intersection of Big Data and Survey Science, on November 4-6 in Utrecht, Netherlands. The session proposal and its four papers are provided in the new documents below.
Following a request from Christian Ruiz (a project collaborator) on analyzing mobile phone data to see whether adopted COVID-19 measures are holding, numerous replies on the topic was received from ML project members and shared on the Covid-19 response : Use of mobile phone data ML Forum page.

Next Steps

Complete pilot study reports and upload them on the public site
Complete a draft of the theme reports (Coding&Classification; Edit&Imputation; Imagery)
Conduct activities on WP3 addressing integration challenges

Risks and Issues

Issue	Mitigation

Input Privacy-preserving Techniques

Progress

We are still looking for an in-kind project manager. Due to this and lack of human resources at UNECE, the project is on hold.

Next Steps

Risks and Issues

Issue	Mitigation

Image result for input privacy-preserving techniques

News from the Groups
Blue-skies Thinking	Identifying Topics/Opportunities	IN PROGRESS	The group is currently focusing on the selected topics. The remaining topics will be reviewed at the May webex call. The BSTN will also consider mid-term responses to Covid19. No country or colleague has submitted new ideas.
	Follow-up selected topics	IN PROGRESS	The Synthetic Data Sets group, lead by Kate Burnett-Isaacs from Stats Canada, had their first meeting and identified two areas where sub-groups will be working on: Methods and tools for producing synthetic data Quality indicators and communication The group now has 20 members and will regularly meet through webex. The Local Level Data-driven Decision making Support group, lead by Branko Josipovic (SORS), reported the experiences by Statistics Serbia and is now looking for other countries to join and share their experiences. The Statsbots group, lead by Eric Anvar (OECD): work is continuing with several countries participating Data Science Lab, lead by Juan Muñoz (INEGI): has further specified the scope. It is looking at the Machine Learning Project and Synthetic Data Sets as potential use/test cases to create a virtual community of experts.
	Other		There were several changes in the group membership. Faiz Alsuhail from Statistics Finland, Carlo Vaccari and Marco di Zio from Istat, Gary Dunnet from Stats NZ and Branko Josipovic (Statistics Serbia) have joined the team.
Capabilities and Communication	Culture Change and Internal Communications Strategy	IN PROGRESS	On hold until the end of May.
	Competencies Training and Development	IN PROGRESS	On hold until the end of May.
	Future of work in the context of Modernisation of the workplace	IN PROGRESS	Countries are starting to share examples of the surveys on working from home, to see how the staff is coming with the current work arrangements.
	Ethical leadership	IN PROGRESS
	Followup Strategic Communication Framework	IN PROGRESS	Prepared on-line publication on Strategic Communication Framework, based on the outputs produced by the Strategic Communication Projects, Phase I and Phase II. Also includes links to the examples of crisis communications from the countries, Covid-19 related communications. Also linked to the COVID-19 wiki of the Statistical Division.
	Training of staff in communication	IN PROGRESS	On hold until the end of May.
	HRMT Workshop	IN PROGRESS	On hold until the end of May.
	Other
Supporting Standards	Linking GSBPM and GSIM	IN PROGRESS	The task team has produced the mapping for phases 2 - Design, 4-Collect and 6-Analyse of the GSBPM. Totally, 12 sub-processes have been mapped by one or two countries. In 2019, the team completed phase 5 - Process. During the meetings, the mapping is presented and discussed by the task team members. The mapping is then revised according to the feedback received during the meeting. So far, all sub-processes for phase 6 have been discussed and updated. The task team is now discussing the mapping for sub-processes of phase 2. Some new challenges are emerging (especially with phase 2) since more conceptual sub-processes are dealt with. The mapping done so far would be very useful in improving the usability of GSIM.
	Core Ontology for Official Statistics	ON HOLD
	GSBPM for Geospatial data	IN PROGRESS	The task team is still without a chair. The task team has started reviewing the descriptions of sub-processes in the GSBPM to emphasize the aspects related to geospatial information/data. Including concrete examples is considered important to help readers to understand what we mean by geospatial information/data. These examples will be also useful to derive general/high-level descriptions later on. During March and April the task team revised all sub-processes of phase 1 - Identify needs and the first sub-processes of phase 2 - Design of the GSBPM.
	Metadata Glossary	IN PROGRESS	The Metadata Glossary Team has completed its work. There is a Metadata Glossary v1_0 draft version that will be made public both as excel file and wiki page. It contains definitions for the modernisation models GAMSO, GSIM, GSBPM and CSPA. It considers the latest versions of the models. GAMSO and GSBPM did not have a glossary so the task team developed new definitions for the main terms. For GSIM , the Group revised the existing definitions of the GSIM Glossary. When the definitions were not following formal rules or were unclear, suggestions for improvements were posted in a wiki area for the GSIM revision task team. Only definitions compliant with formal rules are included in the Metadata Glossary. CSPA terms that do not follow writing convention are removed from the Metadata Glossary.
	Other	The Supporting Standards Group started a discussion about the possibility to incorporate (or merge with) the Sharing Tool Group. This possibility was already discussed by the Group last year, so it was not new topic to the members. The Supporting Standards Groups thinks it would be good to start a discussion about this now and express their opinions. Miain concerns expressed by the Group were around the status of CSPA and potential activities for CSPA that could foillow under the umbrella of Supporting Standards Group. It was decided to have a joint meeting with Sharing Tools in order to check potential synergy and risks.
Sharing Tools	Digitizing/editing CSPA document & Communication restated CSPA	ON HOLD
	Adding Services to Catalogue	ON HOLD	.
	Integrating Innovations in Architecture and technology	ON HOLD
	Other	The group is still lacking a chair (or co-chairs). UNECE is lacking resources to provide the necessary support to the group