Master's thesis repo of Emil Ehnström

This repository contains scripts that has been used in the master's thesis of Emil Ehnström. The Twitter data was collected by using the tweetsearcher developed by Väisänen et al. (2021).

This repository is built upon tools from Digital Geography Lab's borderspace repository The scripts have then been modified for the thesis and scripts from the original repository, that have not been used in the thesis, have been excluded.

The workflow of the thesis follows:

1. Residents of Finland were identified by the twitter-home-detector-tool. With get_fin_residents.py and all_tweets_from_fin_users.py I collect all the data for the user and uploading a new table to a PostgrSQL server.
2. I run the script multilang_detection.py that is sligthly modified version of twitter_multilangid.py by Tuomas Väisänen and Tuomo Hiippala. The script returns a PostgreSQL database.
3. I use the script lang_profile.py that is based upon this script by Tuomas Väisänen. It reads the database of tweets with and identified language and returns a language profile of the user to a PostgreSQL database. I also use the get_lang_summary.py for analysing the amount of users in each language groups.
4. I run the multiprocessing script line.py to create the movement for each language group. The script is based upon works of Håvard Aagesen and Tuomas Väisänen. The lines are later visualised in QGIS.
5. I run a similar script called municipality.py to create movement lines within Finland.
6. To deterimne the diversity I use get_richness_ecopy.py and calculate the Shannon entropy and Simpson index of the unique country travel diversity.
7. For statistics I use cb_summary.py, continents_summary.py, municipality_for_excel.py, and municipality_summary together with Excel.
8. For further analyses in SPSS I use group_comparison.py to get a proper .csv file.
9. Some additional scripts were used for plottin, table writing and other minor operations.

Sources: Väisänen, T., S. Sirkiä, T. Hiippala, O. Järv & T. Toivonen (2021) tweetsearcher: A Python tool for downloading tweets for academic research. DOI: 10.5281/zenodo.4767170