Exploring the distribution of accessibility by public transport using spatial analysis : a case study for retail concentrations and public hospitals in Athens

Sammanfattning: This thesis studies the distribution of accessibility by public transport to retail concentrations and public hospitals in Athens, using spatial autocorrelation analysis based on Local Indicators of Spatial Autocorrelation (LISA) method and Geographically Weighted Regression (GWR). For this reason, the study area was divided in a 300x300 meter square grid of points representing the location points of the study. For each point, the travel time to each of the retail concentration and public hospital location was estimated using Google Directions API. The accessibility of each point was then calculated as the percentage of reachable destinations for each land use group within 45 minutes. The results where then aggregated to zip code level, where along with data regarding the average annual zip code income, population density and destination from Athens’ central business district formed the final dataset. The analysis that followed suggests that there is a cluster of high accessibility in the city center and a cluster of low accessibility in the outer suburbs. In addition, a middle zone of insignificant clustering is located between the two clusters. The chosen urban structure and socioeconomic variables that were used for the geographically weighted regression have significant effects except for the case of population density on public transport accessibility. More specifically, distance from the central business district (CBD) is found to be negatively correlated with accessibility to both retail concentrations and hospitals. On the other hand, annual average income seems to be positively correlated with accessibility to both destinations. Finally, population density has a positive correlation with only retail concentrations. In addition, the analysis indicated that there may be more unknown factors affecting accessibility to retail concentrations in the city center than the periphery.