October 22, 2018

My Experience at Geohackweek 2018

Tianqi Zou

Right before the beginning of this quarter, my first at the UW, I attended the third annual Geohackweek workshop. Hosted by the University of Washington’s eScience Institute, Geohackweek 2018 was a five-day hackweek on geospatial data science, where participants came from all over the world with diverse backgrounds, academic and working status. It was a very rewarding experience, and so inspiring to me, a first year PhD student, who just started the learning and research journey in urban planning.

(Participants of Geohackweek 2018 -by Rachael Murray)

The hackweek was divided into morning sessions and afternoon sessions. In the morning, we engaged in interactive lectures on a wide range of analytic methods and tools for geospatial datasets. We had tutorials on python programming, Github and version control, vector and raster tools, machine learning, Google Earth Engine, and data visualization. Among all the lectures, I found Github and version control, which allows a group of researchers to work on their respective part of a project without disturbing the main chunk of code, and data visualization both programmatic and graphic for geospatial data, were the most useful ones for my current stage of learning and research. My interest in machine learning was also piqued, and I am now seeking more opportunities to learn the methods and integrate it with my future research. All of the tutorials were recorded, and both lecture materials and videos are open and available at https://geohackweek.github.io/schedule.html.

(A morning tutorial session -by Rachael Murray)

 

In the afternoons, we worked on group projects, to put what we had learned in the morning sessions into practice. Basically, all the projects were brought by some participants who had a relatively mature idea of the problem they were going to solve and had a set of data available to work on. Then the organizers facilitated all the participants to form groups with projects they were interested in, and each project group was provided with a data science support lead from the pool of geohack instructors. After hearing several peers pitching their idea, I joined the “Urbanopia” group, led by Lucie Burgess, PhD candidate at King’s College London, working on exploring the relationship between mental health and the built and social urban environment.

Group work -by Rachael Murray

Eddie, Lucie, and me after group presentation

 

Our project was derived from “The Urban Mind” dataset, containing data on mental state in relation to perceptions of the built and social environment. Observations are geotagged which allows context to be developed from the geotagged points. During the hackweek we explored two main problems: 1. The relationship between perceptions of planning inclusivity and street features (such as the availability of street seating, pedestrian crossings, street trees) from OpenStreetMap (OSM) data; 2. The relationship between perceptions of greenness, actual greenness and mental state using NVDI raster data. We broke into two subgroups, and Eddie, Ian, and myself were working on the first problem. Our approach was to convert OSM data from geojson format to geopandas, and extract data on specific tags such as street features, and then develop analytical models to correlate these data with mental states from participants at geotagged locations. Through the week, I got more familiar with python coding and Github, and used it to collaborate with my team. Also, after figuring out the structure of OSM data, I was able to load data from OSM. A week is not enough for us to achieve all our objectives for the project, but now we are still in touch with each other and seeking to contribute to the research.

The week was intensive, and I felt fully loaded of new information and knowledge every day, but the learning and working process was efficient with a welcoming and supportive environment created by my peers and all the mentors. I encourage anyone who wants to gain knowledge and develop skills on geospatial data to apply to the workshop, and enjoy the week of visiting Seattle and networking, as there is also chances to get travel funding from the organizers. Moreover, based on my experience, for future hackweeks, I would suggest the organizers to prepare some “assignments” on each tutorial every day to help participants digest and practice, as the group projects generally do not include every aspects of lectures introduced in the morning sessions. And for people who get accepted to the event and decide to attend, I would suggest you come with a preparation of python programming, and a project idea. Even better would be to have a project on hand, where you need help specifically on analyzing geospatial data. More importantly, I encourage all participants to make full commitment to the workshop through the whole week, which is hard, but is essential not only for you to gain skills consistently and be responsible to your group, but also for the organizers to better manage the events and get more effective feedback.