6 lessons data providers learned during the TWIST hackdays
At the end of August around 80 participants gathered over a weekend at the University of Zurich for the TWIST hackdays which were carried out under the motto “Help us unmask twisted truths with statistics”. The aim was to work with open data that cover the subjects of statistics and emotions, evidence based decision making and fake news.
The event was co-organized by the Zurich R User Group, the Wikidata Community, the Open Network Infrastructure Association and the Statistical Office of the City and of the Canton of Zurich as a pre-event of the Swiss Statistics Meeting.
Data users and providers shared valuable feedback
In a survey conducted immediately after the event, 90% of the participants reported back that they would attend another TWIST event in future. As organizers, we are very happy with that result. However we think that additional critical self-reflection deepens our understanding of how data users and providers experienced the event.
We would like to share 6 lessons we learned from users’ feedbacks and elaborate on how we want to improve the provision of open data in future.
1. Documentation is key
Users lose time and motivation when dealing with poor documentation, or a complete lack thereof. Thus, the provided datasets should be documented in a way that users with different backgrounds have fewer troubles in accessing and understanding them. This is especially important when either the data access or the understanding of the dataset is not straightforward.
2. Involve the experts
Some datasets require very specialized knowledge that may not be acquired in one weekend. In such cases the presence of a data provider makes a big difference. If a data provider cannot be present personally, organizers should make sure that their expertise can be found among the participants. This could be achieved by advertising the event to experts in those specific topics.
3. Disseminating the results
Several teams developed interactive apps and suggested to make them available for use. To facilitate that, it could be beneficial to offer possibilities to deploy developed applications.
4. Get your data ready
Data providers should prepare their data in a meaningful way. They should make sure that hackday participants do not feel as if they are being used as cheap labour. Also, due to data protection issues, data is often aggregated. This makes it difficult to analyze the data more thoroughly. To ensure comparability of aggregated datasets, the levels of aggregation should be consistent (e.g. hexbins cannot be compared with rasters).
5. Set clear goals
Finding a meaningful goal can be difficult, especially when the provided dataset is complicated. To ensure that hackday participants find an interesting challenge to work on, it can be helpful if data providers present a list of potential goals.
6. Let’s not reinvent the wheel
It is a good idea to share code snippets along with data and documentation. This way, hackday participants can save time and focus on answering actual questions.
What are your lessons learned?
What has been your learning experience as data users and providers while participating at an open data hackday? Do you have additional remarks? We are happy to hear from you in order to improve our work in future.
Thank you for your comments below, your feedbacks on Twitter or via E-Mail.
