Creating a tool for the Indian Disaster Resource Network, Kerala.
This post describes the Idea, process, and the visualization of the project as a part of the Interactive Data Visualization course at IDC IIT Bombay under Prof. Venkatesh Rajamanickam @infodesignlab .
There are substantial precedents of disaster India has faced over the few years, and with the more recent Kerala Floods, this shows that we must radically revise our disaster planning and implementation policies that are in place.
But this takes time, so I decided to look at the existing plans laid out ( very few of them in fact exist). In the Process, the IDRN (Indian Disaster Resource Network) came up, a nation-wide electronic inventory of resources that enlists equipment and human resources, collated from districts, states and national level line departments and agencies.
The Platform claims to be “A web-based platform, for managing the inventory of equipment, skilled human resources, and critical supplies for emergency response.”
The primary focus of IDRN portal is to enable the decision makers to find answers on the availability of equipment and human resources required to combat any emergency situation. This database will also enable them to assess the level of preparedness for specific disasters.
Personally, I don’t believe this web application can help make better decisions.
The Platform isn’t as accessible as it should be, and looking at large data that have documents will not help in times of disaster, moving resources across networks requires visualizing these resources, their quantity, their location, and additional details. And yes, I took up this problem as part of the Data Viz Course at IDC @infodesignlab.
The Idea
To start the project, looking at the resources IDRN allocated for the state of Kerala. It was evident that these could have helped in the management of disasters. But, only if decision makers can get an overview of how they are spread across districts and places.
The Tool should help them visualize the Disaster resource network of Kerala.
Aka, A GeoViz Project! I looked no further than Opensource tools (For the Love of Opensource) available to help me along the way! Rasagy Sharma’s workshop on Geodata viz as part of the course, helped me get started with Mapbox studio and Mapbox Gl.
Tools used were:
Mapbox GL JS, It is a JavaScript library that uses Mapbox GL to render interactive maps. It’s an open source library that’s free to use. (Creating the Interactive map)
Google Sheets and their Add-ons. (Cleaning up the data)
Process
Sourcing and cleaning data
Sourced data from the IDRN(Indian Disaster Resource Network) for the 14 districts of Kerala. (http://www.idrn.gov.in/countryquerypublic.asp)
Deriving useful data
Cleaning up the data, getting them into sheets. The data from the reports were taken onto the spreadsheet and cleaned up for the relevant type visualization, looking at GeoViz specific. Sorting the data for each of the 14 districts first and them together on one final sheet.
It’s difficult to geocode each of these locations, here is where GeoCode by awesome table worked like a charm. But cross-checking them took a while. (errors still exist). Tip: Split each address item on each column, and use geocode to get the full address, works way better, results more accurate.
Converting that data into .Geojson problems
Mapbox works best with the .Geojson format, encoding the data takes a while to get used to. After several attempts, from a python code to online converters. This one (http://www.convertcsv.com/csv-to-geojson.htm) helped me along the way. Though errors were persistent, majority of the dataset did come through, but errors still exist (still refining the dataset) and they needed to be looked at individually, will get this sorted out hopefully.
Creating the GeoViz tool
Before narrowing down to Mapbox Gl js, tools like tableau were really interesting but they weren’t as flexible as mapbox gl.
After looking at the dataset, deciding the filter type was essential to the project. With mapbox gl library, you can either toggle a list or search via a filter input. toggle list mode vs filter mode ( Decision had to be made: trial and error method).
Check out some of their examples here.
Too many variables. Stuck to the input filter, though not entirely convinced this is the best way possible, still looking for better ways out there!
Suggestion, sub-grouping the categories of resources might help, in this aspect and toggling the list interaction might work out. But, filtering data on the go was more intuitive. The tool also filters data by zooming into districts and displaying the specific resource of nearby districts as well.
The Tool
Here is the link to the Tool, It’s a work in progress, the dataset needs to be refined and encoded, additional details for each of the resources exist in the dataset but have to work on a better way to showcase them on the popup (Shouldn’t take up too much screen retail). There are bugs, would get them fixed. Hopefully soon.
All the datasets and files are here on my GitHub repo!
I’d love to hear your feedback on this!
Cheers!