Identifying home locations in human mobility data: an open-source R package for comparison and reproducibility
This repository contains the data and code for our paper:
Identifying home locations in human mobility data: an open-source R package for comparison and reproducibilityy. International Journal of Geographical Information Science (IJGIS)preprint
This repository contains all the data and code needed to reproduce the
results and figures in our paper. Although we are not able to share the
raw Twitter data publicly, the aggregated and de-identified Twitter
dataset is found in analysis/data/derived_data/. The steps taken to
produce this aggregated dataset can be found in:
- analysis/00a-de-identification.Rmd: Workflow of de-identification approach
The steps needed to identify home locations for users can be found in:
- analysis/00b-identifying-home-locations.Rmd:
Identifying home locations for users by using four built-in
‘recipes’ in
homelocatorpackage
The steps needed to recreate the figures found in the paper can be found in:
- analysis/01-figures-spatial-comparison-of-four-approaches.Rmd: Geographical distributions of inferred home locations by four different approaches
- analysis/02-figures-spatial-view-of-same-hm-users.Rmd: Geographical distributions of tweets, unique users, inferred homes and actual residents in Singapore (2015)
- analysis/03-figures-correlation.Rmd: The correlation between the normalized number of inferred residents and the normalized number of actual residents in Singapore (2015)
Additionally, this repository also includes supplementary documents for spatial sensitivity test and comparison between raw and de-identified Twitter datasets. As we are not able to share the raw Twitter dataset, we only show the detailed analysis procedures and results.
The steps needed to conduct spatial sensitivity test can be found in:
- analysis/04-appendix-sensitivity-analysis-on-grid-sizes.Rmd: Spatial sensitivity test on different grid cell sizes
The steps needed to compare raw and de-identified Twitter datasets can be found in:
- analysis/05-appendix-comparison-between-raw-and-deidentified-datasets.Rmd: Comparison between raw and de-identified Twitter datasets
Text + figures and data: CC-BY-4.0
Code: See the DESCRIPTION file