Determinants of Airbnb prices in European cities: A spatial econometrics approach (Supplementary Material)

This repository contains supplementary materials for the article:

Determinants of Airbnb prices in European cities:  A spatial econometrics approach

(DOI: https://doi.org/10.1016/j.tourman.2021.104319)

The materials include the used datasets and Python scripts for spatial regression models.

Datasets

For each city two files are provided: data for weekday and weekend offers

The columns are as following:

• realSum: the full price of accommodation for two people and two nights in EUR
• room_type: the type of the accommodation 
• room_shared: dummy variable for shared rooms
• room_private: dummy variable for private rooms
• person_capacity: the maximum number of guests 
• host_is_superhost: dummy variable for superhost status
• multi: dummy variable if the listing belongs to hosts with 2-4 offers
• biz: dummy variable if the listing belongs to hosts with more than 4 offers
• cleanliness_rating: cleanliness rating
• guest_satisfaction_overall: overall rating of the listing
• bedrooms: number of bedrooms (0 for studios)
• dist: distance from city centre in km
• metro_dist: distance from nearest metro station in km
• attr_index: attraction index of the listing location
• attr_index_norm: normalised attraction index (0-100)
• rest_index: restaurant index of the listing location
• attr_index_norm: normalised restaurant index (0-100)
• lng: longitude of the listing location
• lat: latitude of the listing location

Programming Scripts

In this repository you will find a script for spatial regressions in Python using PySAL (models_robust.py).

The codes cover the following regression models:

• OLS
• SLX (lagged_x)
• SAR (lagged_y)
• SDM (lagged_x_y)
• SEM (lagged_e)
• SDEM (lagged_e_x)

Main parameters:

• cities - list of cities from the dataset to be included in the analysis
• Robust=False: calculate the OLS, SLX, SAR and SDM regressions with W (weight matrix) based on 10 closest neighbours
• Robust=True: calculate all regression models with different specifications of W
• direct_indirect=True: calculate the direct and indirect effects (based on Golgher, A. B., & Voss, P. R. (2016). How to Interpret the Coefficients of Spatial Models: Spillovers, Direct and Indirect Effects. Spatial Demography (Vol. 4). https://doi.org/10.1007/s40980-015-0016-y)

Key functions:

• create_weights - defines the W specification
• write_stats - calculates's Moran's I and Geary's C
• direct - calculates the direct effect of the variable
• indirect - calculates the indirect effect
• coord - sets the coordinate refence system (CRS) appropriate to the analysed city
• total_results calculates the regressions
• the coordinates are projected from GPS (epsg:4326) to the local CRS (km_lat, km_lon)
• all regressions are saved as formatted txt table
• the results can be also saved as csv table