Month: December 2014

This. Just this: local mappers made more changes to the map of the Philippines during Typhoon Ruby than anyone else in the world (by a very very big margin). Anyone who doesn’t believe in the strength of local people to build their own resilience should look very, very hard at these numbers.

Ruby’s all over now for the mappers – DHN is de-activated, everyone’s gone back to work.  There’s still a lot of work to do on the cleanup: MarkC mentioned 35000+ houses destroyed and 200000 people without shelter, and there will still be OSM mapping to do for that.  This weekend Celina’s running a “train the trainers” OSM event in Manila: if you’re one of the people who created the figures above, please please go and help spread your skills further!

I’ve been talking to the HDX team for some time now (well, since before HDX was a thing, but then so have many of us).  HDX is a data repository for humanitarian data: basically, it’s a place to put machine- and human-readable datasets so that other people can use them too.

Ushahidi tools (Ushahidi platform instances, Crowdmap instances) often have datasets in them that could be useful to other people, so part of the conversation has been about how to share data from Ushahidi sites, both on the HDX site and in the HXL humanitarian standard.

Ushahidi CSV to HDX CSV

First let’s look at how to share the CSV file that Ushahidi creates when you click on the “download reports” button. Before you do this, please, please, please read my post about mitigating potential risks to people from sharing your Ushahidi data.

Converting that CSV into HXL format is pretty easy: it’s as simple as adding a row to your CSV file.  Under the heading row (“#”, “INCIDENT TITLE”, ”INCIDENT DATE”, “LOCATION”, “DESCRIPTION”, “CATEGORY”, “APPROVED”, “VERIFIED”, “LATITUDE”, ”LONGITUDE”), add another row with the HXL tags (#x_uniqueid, #x_reporttitle, #report_date, #loc, #x_description, #x_category, #x_approved, #x_verified, #lat_deg, #lon_deg).  Erm… that’s it.

Loading the CSV file into HDX is also pretty easy: you go to HDX, create an account (and an organisation, if you need to), then click on the ‘upload data’ button and follow the instructions.

Ushahidi API to HDX CSV

I did this recently for some Ushahidi location data (I used the API to pull out all latitude-longitude-placename rows from the Ushahidi instances used for the last Philippines typhoons, then wrote them to a CSV which I cleaned up a bit before posting manually to HDX).

If you can write (or run) Python, you can access a Ushahidi Platform API and get all the reports (or locations or categories etc) from an Ushahidi Platform instance – IF that instance has turned on its API (most have). This allows you to do more complex things, like get data from multiple Ushahidi instances, process it a bit (e.g. removing all those repeat locations) or merge it with other data (e.g. Pcodes).

Geeky bit (non-coders, look away now): This API to HDX transfer could be done as a 1-click button on the reports page, if we add an “HDX” plugin to the Ushahidi Platform.  The design for the plugin is basically: a) allow users to add their HDX settings using the plugin’s “settings” page, and b) add a  button to the reports page, and use the existing Ushahidi CSV creation code to create a CSV file, then push that file through the HDX API using the settings that the user’s already added. This would be a pretty sweet Ushahidi plugin for a community coder to write: it’s well-defined, and most of the code needed is already there.

Ushahidi API to HDX API call

HDX not only allows you to upload datasets, it also allows you to add a link to an API, so HDX sees the current information from a site (not a snapshot that you’ve turned into a CSV file).  You can do this by adding one of the Ushahidi API calls to the HDX “link” field, e.g. if your site is called www.yoursite.com, 

• https://www.yoursite.com/api?task=incidents&limit=1000 – for reports (NB this only grabs the first 1000 reports)
• https://www.yoursite.com/api?task=categories – for the categories list
• https://www.yoursite.com/api?task=locations – for the list of locations

Again, please please please read my notes about mitigating potential risks to people from sharing your Ushahidi data before doing any of these.

HDX to Ushahidi

The HDX datastores include files in CSV, Json, XML and Shapefile formats.  You can import Ushahidi reports data from HDX into an Ushahidi instance (but be sure you have permission to do this before you do); you can also upload shapefiles as layers (converted to WMS or KML format: the ogr2ogr tool can do this for you) into an Ushahidi site.

Helping out with HDX and Ushahidi

HDX is an open-source project (https://github.com/OCHA-DAP/): if you’re a Python coder, please consider helping out with its  codebase.

The HDX team are also working on unlocking useful data that’s still in PDFs. If you want a small task to help them with that, try this: unlocking humanitarian planning data.

The Ushahidi Platform is also open-source (http://github.com/ushahidi): if you’re a Php coder, please consider helping out there too.  The plugin specification above could be a nice place to start…

Cross-posted from http://icanhazdatascience.blogspot.com/.

I recently taught a coding course for international development students (yes guys, I’m still marking your assignments). But teaching coding isn’t useful in itself – it’s like teaching someone a new language without understanding what they’d want to communicate in it – so woven throughout the course was a design exercise, so students would end the course with a design for a system they wanted to build, and a tangible use for their new skills.

We focussed on user experience (UX): the process of designing a system from the user’s point of view, rather than the coder’s assumptions about what the user might like. And specifically on a set of UX tools that together define a system:

• personas
• user journeys
• sitemaps
• wireframes
• graphic designs
• user stories

I pointed students at two great resources on this: the beginner’s guide to UX and UX Apprentice, but here’s a quick run-down on the basics.

Design tools

If you’re doing visual designs (wireframes, site maps etc), pen and paper is both free and easy to edit, but not so good at storing and sharing edits between people electronically. Better tools for that are Balsamiq (free for students) and Pencil (less powerful, but free for everyone).  And Trello is a good tool for user stories.

Top-level design

The top-level UX elements are personas, user journeys and sitemaps.

Personas are examples of each type of user you expect to interact with your system. They’re used to get to know the people who will use your system, understand the problem that they’re trying to solve with it, and how people already solve that problem.

User journeys (aka scenarios: they’re almost the same thing) describe how the user will solve their problem.  A user journey contains elements including

• Context – where?
• Progression – what are the steps in this journey?
• Devices – mobile? laptop? smartphone?
• Functionality – what do they expect?
• Emotion – angry? tired? scared? time-pressured

A sitemap lists all the pages on your site, and how they’re related to each other. Sitemaps can be generated by a “card sort”, where you write all your system functions (or user stories) on post-its, then sort them into groups.

Page design

The UX elements for individual pages on your site are wireframes, mockups and graphics.

A wireframe is a deliberately rough sketch of what’s on the page. It’s deliberately rough to force your stakeholders (users and customers) to concentrate on *what’s* on the page, not what colour it is, which images you’re using etc.  You might see the text “lorum ipsum” on mockups: this is standard ‘nonsense’ text used by designers, and can be found on sites like http://www.lipsum.com/.

A mockup is more detailed, and used as a proxy for the completed page in exercises like talking potential users through the functions of a site. Clickable mockups are good for this (Balsamiq wireframes are clickable, as in you can click a button on one wireframe and the code will display the page that the button goes to), as are printouts of mockups if you don’t have a machine available.

Graphics are the look-and-feel of your site, its branding and things that make it seem ‘familiar’ to your users.  This includes the color palette you use, fonts, shapes and images (icons, header images, backgrounds etc). Not everyone who codes can design graphics, so you main choice is either to find someone who can design and build all the graphic elements, or use packages of ready-made graphical elements like Bootstrap or Foundation with tools like DivShot.

When you design your pages, it helps to look at similar websites, and ask yourself what you do and don’t like about them, then consider that in your designs.

Back-end design

The above has been all about designing the front-end of a system, e.g. the things that a user can see and the interactions the user has with them.  Back-end design (e.g. algorithm design) is a different beastie, and one to be covered later.

Project management design

Design elements used for communicating with a coding team (including back-end coders) include user stories, kanban and minimum viable products.  These are the foundation of “agile development”, a process designed around building parts of each system quickly in “sprints” (short cycles of coding and feedback) so you can get user feedback on them (and adapt your design to that user feedback) as you go.

User stories are used for planning code sprints, and look like this: as a <role> I want to <goal> in order to <benefit>

Kanban charts are those cards (or post-its) on the wall that you see in every photo of a modern software company.  They usually have cards organized into 5 columns: backlog, ready, coding, testing and done, with a user story or task on each card. An agile project manager (“scrum master”) will pick the most important user stories on their list, and add them to the kanban chart at the start of each code sprint. As programmers write code for each user story, the story will move from left to right across the chart until it ends up in the “done” column.

A minimum viable product (MVP) is the smallest system that you can build and still get useful feedback on. You build this, get feedback, adapt your design and repeat.

That’s it: a basic run-through of design tools for coding.  Now go out there and design something!

I have to go back to the day job today, so here are some places you (yes, you!) can help:

• OpenStreetMap tasks list
• micromappers clickers
• Digital Humanitarian Network (find a group that sings to your heart and join it: OSM, SBTF, GIScorps, Translators Without Borders are all active for Ruby).

 

 

I’ve been thinking lately about how open data and community data fit together.  Actually, I’ve been thinking about it for a long time – since we launched OpenCrisis to try to get more data tools and ideas into the hands of crisismappers, and started the long work of trying to archive and share as much mapping data and information as we could. Here’s some first thoughts on restarting this work.

Some of the big questions that came up then, and continue to come up, are about ownership, risk and responsibility.  For instance:

• Ownership. If a community of people add reports to a site, and that site also sucks in data from social media, who owns that data?  That question’s already been asked and answered in many open data and social media sites, often involving much work and lost data as licenses are transferred (see OpenStreetMap’s license moves, for example).  Having your mappers sign a contributor agreement, and having a data license statement on any dataset you put out is a good start here.
• Risk. Is something we’ve also always dealt with. The types of information that mappers handle mean you need to do data risk analysis that covers population, mappers, organizations and leads.
• Responsibility. If you make a dataset public, you have a responsibility to, to the best of your knowledge, skills and advice you can obtain, do no harm to the people connected to that dataset. That’s a big responsibility, and one that has to do a balancing act between making data available to people who can do good with it and protecting the data subjects, sources and managers.

This week, I used some Python code to share all the geolocations (latitude, longitude and location name) from 2013 Ushahidi instances for Philippines typhoons.  That was a pretty clear-cut example: it wasn’t a conflict situation, none of the locations were sensitive (e.g. personal homes or addresses of places like rape crisis centers that need to be kept secret), and the data was gleaned from public sites created by large NGOs and open data teams.  Life isn’t always that clear-cut.

Datasets: where and what

Some of the places that mappers hold data are:

• Google spreadsheets (tables of information),
• Skypechats (informal information transfers) and googledocs/emails (e.g. annotated highlights from a day’s Skype discussions),
• OpenStreetMap
• Micromappers data (often visible as google spreadsheets) and
• Ushahidi instances (unfortunately afaik, there weren’t any Ushahidi instances updated for Typhoon Ruby, so I couldn’t compare the two sets of data).

Some of the data collected by those mappers includes:

• Geolocation: Latitude and longitude for each location name in a list.  These are usually either a) found automatically by the platform using a gazetteer like Nominatim, b) input by site administrators, or c) input by the person submitting a direct report.
• Direct reports: Messages sent by reporters or general public via SMS or web form.  These generally have the standard set of Ushahidi report entries (title, description, category list etc), but might also include custom form fields to capture items of specific interest to the map owner.
• Indirect reports: Messages scraped from Twitter, Facebook etc.
• Category lists: The categories that reports are tagged with; these lists are usually created by the site administrator.
• API data: data input into a platform by another system, using the platform’s API. This includes sensor data, and datasets scraped and aggregated from another platform instance.
• Media: Images, video, audio files added to a report by the reporter or an administrator.

Not all of this data is owned by the site owner.  For example, third party data (e.g. Twitter messages) has restrictions on storage and ownership that even with the original sender’s permission could make it illegal for you to distribute or keep on your site.

Who and why?

Open data is, by its nature, open, and it’s difficult to anticipate all the uses that people have for a dataset you release.  Some examples from experience are:

• Academics – analysis of social media use, group dynamics etc.
• People in your organization – for lessons learned reports, for illustration, for visualizations, for analysis of things like report tempo (how frequently did information arrive for processing, translation etc)
• Data subjects – to check veracity of data about them, and to ask for data removal (sometimes under laws designed for this purpose, e.g. EU privacy laws). I haven’t seen this happen yet, but I really really want it to.

If you release a dataset to anyone, you have to assume a risk that that dataset will make its way into the public domain. We’ve seen too many instances of datasets that should have been kept private making it into the public domain (and, to be fair, also instances of datasets that should have become public, and datasets that have been carefully pruned being criticized for release too).  Always have the possibility of accidental release in mind when you assess the risks of opening up data.

How?

Sharing data shouldn’t just be about clicking on a “share” button. There are processes to think about, and risk analysis to do:

• Ethical process: Assessing the potential risks in sharing a dataset; selecting which data you should and should not share. Alway think what the potential harms from sharing information from the deployment is, versus the potential good.  If you’re not sure, don’t share, but if you’ve checked, cleaned, double-checked and the risk is minimal (and ethical: you’re working with other people’s information here), seriously consider it. If it’s come from a personal source (SMS, email etc), check it. At least twice. I generally do a manual investigation by someone who already has access to the deployment dataset first, with them weeding out all the obvious PII and worrisome data points, then ask someone local to the deployment area to do a manual investigation for problems that aren’t obvious to people outside the area (see under: Homs bakery locations).
• Legal process: choosing who to share with, writing nondisclosure agreements, academic ethics agreements etc.  You might want to share data that’s been in the media because it’s already out there, but you could find yourself in interesting legal territory if you do (see under: GDELT). In some countries, slander and libel laws could also be a consideration.
• Physical process: where to put cleaned data, how to make it available.  There are many “data warehouses” which specialise in hosting datasets.   Data warehouses include the Humanitarian Data Exchange (HDX), which specialises in disaster-related data.  You can also share data by making it public on an Ushahidi instance (e.g. crowdmap), or by making data available to people on request. See crowdmap.com’s api and csv public download button.

Some of the things I look for on a first pass include:

• Identification of reports and subjects: Phone numbers, email addresses, names, personal addresses
• Military information: actions, activities, equipment
• Uncorroborated crime reports: violence, corruption etc that aren’t also supported by local media reports
• Inflammatory statements (these might re-ignite local tensions)
• Veracity:  Are these reports true – or at least, are they supported by external information.

Things that will mess up your day doing this include untranslated sections of text (you’ll need a native speaker or good auto translate software), codes (e.g. what does “41” mean as a message?) and the amount of time it takes to check through every report by hand.  But if you don’t do these things, you’re not doing due diligence on your data, and that needs to happen.

Other questions that you might want to ask (and that could make your checking easier) include:

• How geographically accurate does your data release have to be?  E.g. would it be okay/ better to release data at a lower geographical precision (e.g. to town level rather than street)?
• Do you need to release every report?  Most deployments have a lot of junk messages (usually tagged unverified) – remember, the smaller amount of data you release, the less you have to manage (and worry about).
• Would aggregated data match the need of the person asking for data?  e.g. if they just need dates, locations and categories, do you need to release text too?
• Time. You might want to leave time after your deployment for the dataset to be less potentially damaging. When you release a dataset, you should also have a “data retirement” plan in place, detailing whether there’s a last date that you want that data to be available, and a process to archive it and any associated comments on it.

Further reading

There’s a worldwide discussion going on right now about how private individuals, groups and companies can release open data, and on the issues and considerations needed to do that. More places to look include:

• Responsibledata.io page on responsible development data 
• Responsibledata.io resource page
• Ushahidi Data Cleaning Guidelines

All datatypes:

• UNOCHA COD/FOD datasets
  • 2013 Philippines population estimates
  • Philippines airports (see also OurAirports)
• HDX Ruby
• Open Data Philippines
• Whatever we can grab off people as they go past

GIS datatypes:

• Yolanda GeoNode
• Philippines GIS data clearinghouse (WorldBank?)