Machine Learning

Progress

A follow-up was initiated on three of the four topics identified by the BSTN core group and directed to the ML project. One of the follow-ups led to Statistics Norway joining the project to collaborate on the use of ML to code NACE. We await a response from a member of the BSTN core group on another topic. The membership of the project now stands at 32 participants from 13 countries.
The objectives, deliverables and timelines of each work package and pilot studies will be completed by the end of the month. Most countries are preparing the required data. The approach taken in the various studies is to share code and test them each organisations' data, rather that attempt to create common datasets that would be accessible and meaningful to all organisations.
The pilot study on Coding and Classification now includes 7 projects from 6 countries. Statistics Poland has used code from the US-BLS on an ECOICOP classification and achieved an initial accuracy of 88%.
The project manager wrote a short report describing the scope and progress for a meeting of chief statisticians at the CES plenary session in Paris. Two members will be introducing the project at the ModernStats workshop in Geneva.
Discussions are taking place to set the date and location of a second face-to-face sprint. A decision will soon be taken.

Next Steps

Set up an environment to share code (refer to risks below)
Conduct the various studies, regularly monitor progress and gather results.
Organize the next face-to-face meeting.

Risks and Issues

Issue

Mitigation

We need to setup immediately an environment to share and manage machine learning code. GitHub is the preferred option. Given the number of participants, we cannot use a free account. One needs to be purchased. Discussions on this are taking place with the UNECE.

For now, we are testing a temporary approach using Confluence. This will also be necessary until a permanent solution is acquired and setup.

Some organisations are very interested in learning how to analyze imagery data. Mexico will be conducting a pilot project using Landsat open data to measure and monitor urban density. Other organisations could learn to use satellite data by collaborating with Mexico. Such a collaboration would be facilitated if they can acquire satellite Analysis Ready Dataset (ARD). These are not produced by NSOs.

This in not a risk, but rather a challenge. Resolving it would further augment the impact of the ML project. Exchanges are taking place between ABS, INEGI and third party data processors. We raise this as a "heads-up" in case we need the support of the EB to move this issue forward.

Strategic Communication Framework phase 2

Progress

We had an excellent two day sprint in Gdańsk Poland on June 10 and 11. A special thank you to Statistics Poland for being such gracious hosts to both the sprint and the Workshop on Dissemination and Communication.

Work is progressing well. We developed some content and have excellent work plans for the remaining work related to work packages 1 (Mission, vision and values / Staff Engagement) and 2 (Stakeholder Engagement). We also have sufficient and dedicated participants willing to contribute to the work required to complete these two activities. During the Workshop of Dissemination and Communication, we ceased the opportunity to promote the results of Phase 1 and present the work plans for Phase 2. This resulted in greater awareness amongst the broader international communications community. It also resulted in the addition of a new member to the project team from Oman.

As for work package 3 (National Data Strategies), we await the outcome of the CES Seminar later this month. In the meantime, we have begun to gather an inventory of countries engaged in national data strategies, examples of national data strategies and have commitment from a couple of countries to produce case studies on their experience to date. In addition, we have identified a strategy that we believe can best contribute to this exercise at this time. We have begun to develop content regarding the considerations NSIs need to explore to ensure public acceptance of this new business model.

Next Steps

Work will continue over the summer on work packages 1 and 2. We will await the results of the CES Seminar in a couple of weeks to determine whether adjustments need to be made to work package 3.

Eurostat has agreed to act as an editor on the final draft products and INEGI has agreed to take on the role of producing our final products. These commitments are very much appreciated by the project team members.

Risks and Issues

As we began work late this year, the timetable is very tight to deliver a solid draft product by the HLG Workshop in November. However, we believe we have established a solid workplan and have the necessary resources that will allow us to deliver a quality product on time.

News from the Groups
Blue-skies Thinking	Identifying Topics/Opportunities	IN PROGRESS	14 Topics were identified and the next steps were decided on. Project Proposals submitted in 2018 that were not selected, will now be considered by the core group. The group is also considering to organize an IT Strategy meeting of CIOs.
Blue-skies Thinking	Follow-up selected topics	IN PROGRESS	The following two topics will at first be discussed in more detail. A position paper will be prepared for each: Synthetic data sets: Statistics Canada (lead), ABS, Istat, and Statistics Netherlands Secure Multi-party Computation: ABS (lead), Stats Canada, ONS and Statistics Netherlands INEGI, CSO Ireland and Statistics Netherlands will do a further refinement on the Data Science Lab proposal.
Developing Organisational Capability	Skills and Capability Framework	IN PROGRESS	We have got first initial draft of paper on importance of complementary skills. It was decided that day before Culture workshop in September we will have sprint session on it. All DOC group members are obliged to develop this paper by their experiences up to half of August.
	Promotion Forum	IN PROGRESS	We prepared new leaflet for CES in June to promote our activities.
	Setting vision in NSOs	IN PROGRESS	We are waiting for the results of Group on Communication.
	Other	The Organising Committe of Culture Workshop had a webex call. So far we have got around 10 papers so we need to fill gap, especially in the area of vision and mission. On 26th of June will be next call. Our goal is to encourage all interested people to present their experiences in given areas. So we would like to ask EB members to share this information with thier colleagues.
Supporting Standards	Linking GSBPM and GSIM	IN PROGRESS	The task team is meeting regularly every three weeks. A template for the mapping has been agreed. The mapping is being done at two different levels of GSIM: a) a more conceptual level, corresponding to the specification level in GSIM; b) a less conceptual level, corresponding to the execution level in GSIM. The task team is concentrating on phase 5 of the GSBPM both at a design and implementation level. Last meetings were dedicated to reviewing the mapping and the examples provided by different countries. This implies an harmonisation process, on a one side and a consolidation of the templates, on the other. As already mentioned to the Executive Board meeting in May, the task team will be able to do the mapping for phase 5 and 4 and maybe one additional GSBPM phase. It is likely that the task team will not be able to complete the mapping by November.
	Core Ontology	IN PROGRESS	The development of the core ontology goes on at a steady pace, via virtual meetings and offline exchanges. The construction of the model first focused on the integrated view between GSBPM and GAMSO, with discussions on the connection between the notions of activity and process. More recently, modelization of the statistical organizations and products was undertaken. A first version of the ontology will be available for presentation at the HLG meeting in November and afterwards submitted to public review.
	Alignment GSBPM and GAMSO	IN PROGRESS	The task team has produced a document specifying the activity to be done. The document is regularly updated after each meeting. Agreement has been reached on the content and the structure of the document. The premises of the document were consolidated. The discussion on the overarching processes in GSBPM and their relationships to GAMSO corporate support activities, started with the OP "Quality management".
	Metadata Glossary	IN PROGRESS	The work is proceeding regularly
	Other	The Supporting Standards Group was higly involved in the preparation of the June ModernStats World Workshop. The Workshop will be the occasion to present the ongoing work of the Group and receive feedbacks. Some small group discussions are being organised about topics on "Terminology", "Statistical Activities in GAMSO" and Items on "GSIM".
Sharing Tools	Digitizing/editing CSPA document	IN PROGRESS	First the CSPA Document will be updated and edited. The Application Architecture section has nearly been completely rewritten. The 'Common Statistical Production Architecture' section had some significant changes. Smaller changes were made on: Business Architecture, Information Architecture, Enablers. CSPA V2.0 Overview (New) was added.
	Adding Services to Catalogue	IN PROGRESS	Several new services were added to the Catalogue. During the ModernStats World Workshop we will have interactive activities in which potential services will be identified and some selected will be tried to add to the catalogue.
	Communication restated CSPA	NOT STARTED
	Other	Several members were active with preparations for the ModernStats World Workshop. We'll have several presentations and interactive sessions to promote and advance CSPA.