2020-09-22, 15:05–15:25, Room 1
This talk focuses on the normalization of Twitter users data collected in Istanbul, Turkey to provide a base for certain research topics such as locating emergency meeting points, detecting unexpected events, relieving traffic jams.
A specific location of a Twitter user is obtained from all corresponding tweets in a time interval by applying an iterative algorithm that generates a circle of a 1km radius for each tweet’s location and finds the most -intersected points in the circle. Afterwards, the location of this user is allocated in the municipality provinces in order to compare with the census data collected in the same time interval. With such a comparison, day / night maps (night maps are created by using census data, while day maps by using Tweets), normalized maps acquired by dividing the amount of Twitter users to the population in the corresponding provinces.
To achieve the aforementioned tasks, Tweets are acquired by using GeoTweetDownloader (GTD) in a GIS laboratory of the Istanbul Technical University. Some Exploratory Data Analyses are performed via Matplotlıb, Seaborn and Geopandas. Furthermore, the conversion of geographic coordinates to projected coordinates for creating thematic maps and interactive maps are completed by pyproj, shapely, mplleaflet and folium libraries.