Streetception

How safe or attractive do different people perceive streets to be and why?

In this repository we share the code and data we collected and analysed for the paper "Is it safe to be attractive? Disentangling the influence of streetscape features on the perceived safety and attractiveness of city streets".

The data contain ratings of perceived safety and attractiveness (using a 5-point Likert scale) coming from 403 participants who were asked to virtually navigate city streets in Frankfurt, Germany, through a sequence of street-level images. Moreover it contains their explanations of the ratings (in their own words).

In total we have collected data for 753 locations. In particular:

• 7989 rating pairs of perceived safety and attractiveness
• 19114 keywords used to explain the safety ratings
• 18232 keywords used to explain the attractiveness ratings

Data

A DOI to the data can be found here: https://doi.org/10.5281/zenodo.7712811

Structure of the data

[data_per_location.csv]

geometry_text	path_id	image_point_id	order_id	{safety,attractiveness}_{avg,stddev,difff}	{safety,attractiveness}_reason	{male, female, age_19_39_safety, age_40_59_safety, age_60_plus_safety}_{avg, count}
EPSG:4326	id of the path this location belongs to	id of the image	order of location in the path	Average, Stddev, and difference of the safety and attractiveness ratings collected for this location (number)	Reasons to explain the safety and attractiveness ratings the participants provided (text input)	Number and average ratings of safety and attractiveness the participants provided per gender and age group

[text_attr_peruser.txt, text_unattr_peruser.txt, text_safe_peruser.txt, text_unsafe_peruser.txt]

All reasons participants provided to explain the ratings of locations with an average rating of safety or (attractiveness):

• rating <= 2.5 (text_unattr_peruser.txt, text_unsafe_peruser.txt)
• rating >=3.5 (text_attr_peruser.txt, text_safe_peruser.txt)

Example

Notebooks

Under the folder "notebook" you can find two notebooks

• analysis_experimentation.ipynb: contains code for the analysis of the influence/relation of safety and attractiveness using the participation rating
• word_analysis.ipynb: contains code for the analysis of the text-based input collected from the participants