This post follows up on information shared earlier:
The purpose of the post is to highlight data and technical details for an ongoing project, which will also be the subject of our work at the Open Data for Development Challenge (Montreal). Readers should first familiarize themselves with the general details outlined above.
We start here with questions related to the end-point or final goals, and work our way backwards to details about what we would like to use and experiment with. Please see the link above for reasons why we are undertaking the project.
To summarize , we have a number of objectives: a) experiment with joining different data feeds and sources to shed more light on a signature Canadian foreign aid initiative, b) assess the degree of financial ‘trace-ability’ that is possible, given the current state of data publication around Muskoka-MNCH initiatives, partners involved, and their various reporting, c) contribute ideas towards building a more effective knowledge and information sharing system that taps into results, impact, and monitoring & evaluation data around Muskoka-MNCH initiatives from partners, d) contribute ideas towards enhancement of the IATI standard so that it can more effectively capture signature donor initiatives (for which there is no coding presently) and better leverage results and impact information from project partners.
These interrelated objectives are intentionally broad as we cannot anticipate which aspects we will be able to work on at coding events, as a lot depends on the make up of the teams (which we do not know beforehand). So we purposefully leave it broad so all participants – whether coders, policy, domain, or data experts – have areas they can contribute to.
The remainder of this post further specifies technical objectives, associated data, and known challenges.
Build a comprehensive visualization that tracks Canada’s Muskoka-MNCH commitments and actual expenditures at a granular level.
The visual should be connected to several data sources (see details below). To demonstrate work with ‘live’ data, the visual can connect to Canada’s IATI publication. However this has some shortcomings as far as meeting our needs on Muskoka-MNCH is concerned (see further below). Other options would be to connect to alternative data sources (discussed below), and or demonstrate connection with ‘live’ online versions of our data, such as through Google Fusion Tables.
The visual should allow for ‘real-time’, or periodic updates, as data comes in.
The visual should allow users to navigate across various levels; from aggregate commitments per country, region, or initiative, to expenditure at the project level.
The visual should show “draw down” on project budgets, i.e. it should show expenditures over time, against the total committed amount.
The (suggested) components of the visual include: a map of activities (e.g. aggregated up to the country level); a network graph of implementing partners; and a table that allows users to drill down further for more granular detail. The general idea is for the navigation to be driven by three elements: what (project details); where (map); and who (partners), couched within wider financial commitment and expenditure information. These are merely suggested entry-points. We are well aware from past experience that not all elements may be achieved.
In order to demonstrate ‘joining up’ of multiple data sources and or data feeds, the map could be layered such that project details are shown over a heat-map of the scale of MNCH issues, i.e. health related indicators. Linking up, either live (using the World Bank’s Health Stats API) or with a static dataset, is fine for demonstration.
Eventually we expect to link the visual with data on impacts and results information collected directly by partners in the Canadian MNCH alliance network (their impact tool is currently under development). Some consideration should be given to how data can be fed in ‘real-time’ (not specifically “real-time” as in up to the minute, but updated periodically).
Developing reusable heat-map layers based on MNCH related health indicators would be valuable. Heat-map layers could include data on select MNCH indicators, as well as data on the state of health expenditure (e.g. health $’s per capita) and the composition of health expenditure (e.g. public vs. private).
Please note, we are making all attempts to convert and post all relevant data as Google Fusion Tables (GFT). The data will be discussed with the team in detail.
Comprehensive list of all Canadian Muskoka-MNCH related unique project IDs.Musk-MNCH-UIDs (excel version; CSV and fusion tables may be available). These IDs can be used to query against a range of Canadian open aid data sources, as well as Canada’s (i.e. DFATD’s) IATI publication.
Consolidated customized Muskoka-MNCH data. ConsolidatedMuskoka-MNCH dataset (excel version, large file; to be split in parts and converted as fusion table).
Custom dataset we created joining up Muskoka-MNCH data from our consolidated dataset, as well as other Canadian sources (HPDS and projects browser). Contains unique IDs with project start end dates, fiscal year expenditures, organization and type of organization, and implementing partners, titles and more. MuskMNCH JoinedUp set (Cons-PBw-HPDS)
Please note: the above three datasets have been customized for this project, and contain unique IDs on which they can be joined.
Other aid data:
Canadian (DFATD) IATI data.
Select implementing partner IATI feeds: implementing partners through which Canadian Muskoka-MNCH funding is channeled, that are already publishing to IATI. These include multilateral partners such as UNICEF, UNDP, WFP, UNFPA, the GAVI alliance, the Global Fund, the World Bank
Specifically we would be interested to see if there is any possibility of Muskoka-MNCH funding showing up in their IATI publication, whether as projects, or in their discussion of results related to MNCH. The Muskoka-MNCH baseline calculation method can be used to asses which DAC-CRS codes (bilateral aid) and ratios for multilateral contributions count as Muskoka-MNCH.
Aid Data 3.0: could be useful for DAC-CRS data which otherwise is not available via an API, for IATI data, and for other donor project level data.
Indicator data from Countdown 2015. (optional, the formats are not consistent; some of this data may be available more readily from other sources like WHO or the World Bank, see below).
Google Fusion Tables: while still a beta product, fusion tables can be extremely useful for basic mapping, creating map layers, and for basic data joins and mashups. However there are some important limitations (file sizes, speed, update frequency) that developers should be aware of if they are not already. Fusion tables can be used to run Google Maps, however there can be limitations in terms of functionality, interactivity which can be complicated to overcome.
d3 – Maps: there are several mapping examples in d3 but this provides a good primer from basic to more advanced. d3 maps have also been used to link up “real-time” data, for e.g. on Wikipedia updates around the world, that could serve as inspiration for tools that could be used to get around our need to periodically refresh with new data (real-time data map example).
d3 – network graphics and tables: d3 contains several examples of node-link, hierarchical, collapsible tree layouts, and other options that could be good for network graphics to visualize Muskoka-MNCH partners and links between the same.
ArcGIS: could be used for mapping and map layers.
MapBox: is another powerful mapping toolkit.
Tableau: we use a range of Tableau products in our work. However custom maps and geocoding created using other tools can be brought into the Tableau environment (e.g. as polygon, or shape files). Developers may be able to experiment with bringing custom layers into Tableau, working with the NSI project lead.
Custom CSS, HTML: GIS experts and coders may well be more familiar with other tools. There is always the possibility of designing things using other tools
The (known) Challenges and Work-arounds
IATI data does not contain coding for “Muskoka-MNCH” (to our knowledge).
IATI data (i.e. Canadian, DFATD) does not contain expenditure information (yet), to our knowledge.
We can get around this by using our custom datasets, and cross-referencing the project unique IDs against IATI (and other sources, for e.g. DAC-CRS), as well as other Canadian sources.
The Muskoka-MNCH commitment is actually complicated as the $2.85 billion over 5 years Canadian commitment breaks down into $1.1 billion in “new and additional” spending, and $1.75 billion in ongoing or baseline spending. This again creates challenges in terms of using IATI (e.g. for legacy data). We can get around this again by using the custom datasets provided. Wherein this is simplified and already coded. However it does present a challenge for standards like IATI in that similar issues could come up in the future as other donors announce signature initiatives. Therefore something the IATI standard should give consideration as analysts and viewers tend to be interested in highly publicized aid initiatives, and expect up to date information on the same via open aid data sources.
Questions viewers would like to address using the tool
How much is committed via Muskoka-MNCH, and where?
Is spending on track to meet the commitments; overall at $2.85 billion over 5 years, and specifically at the country level?
Who are the implementing partners? Does Muskoka-MNCH relevant information show up in the IATI publication of implementing partners (those that are publishing to IATI, such as the multilateral agencies noted above)?
Is funding going to countries most in need; as can be seen from the scale of the problem, the trends in addressing the challenges, and availability of resources (i.e. the level and composition of health expenditure)?
What percentage share of any $1 in the ‘supply chain’ of Muskoka-MNCH funding is visible, given the current state of publication, either to IATI or other sources, by official agents and or implementing partners?
What view of future Muskoka-MNCH spending do we get from the data? (e.g. from forward spending plans, and or from project level details).
What sort of results and impact information is available (e.g. what results information is reflected in the IATI data for Muskoka-MNCH projects)?