Gradu #maastopyöräily'stä Pallas-Ylläksellä, Pyhä-Luostossa ja UKK:ssa. Voisivatko urheilusovellusten massadatat auttaa kansallispuistojen suunnittelussa? Mikko Kangasmaa selvitti 👆

BONUS ✨
Random highlights from the dataset:
1) Median time between capturing a photo and uploading it to Flickr is one week.
2) People (or bots) like to upload on round figures (see pic)
3) Normalized by population, Iceland has the most Flickr users in our dataset.

Thank you to #Kone & Mai and Tor Nessling Foundations for supporting this work. A quantitative work like this would not be possible without a robust suite of FOSS tools. My thanks to the maintainers of #QGIS, #pandas, #geopandas, #duckdb, #dask, #statsmodels, #jupyter and many more!

We hope to help future research by sharing the aggregated datasets presented above as figures and processing codes to produce them: buff.ly/qoJeQn8

So, we argue that these trends and biases should be considered if opting for Flickr data. For example, if Flickr is used a proxy does a drop tell about the phenomena or the popularity of Flickr? All research that uses user-generated dataset opportunistically face similar questions.

(5) Finally, we tested the reliability of data acquisition from the platform and found that the API responds inconsistently to repeated queries.

(4) All of this is reflected on who make posts from nature, or protected areas in this case. For example, Europeans and North Americans make about ¾ of the visits to African protected areas on Flickr.

(3) Flickr is clearly a platform of the "Global North": it’s users and use are concentrated in Europe and North America. Over time, an even greater share of use is in Europe.

(2) More of that remaining data is produced by the most active ‘super-users’ of Flickr – top 1 % of users uploaded about 1 / 3 of the photos!