Gudrun Haindlmaier (p)

The provision of pharmaceutical products in Austria is highly regulated and followes the rationale of supplying the residential population area-wide mainly by means of public pharmacies (to some extent in rural areas also by private medical pharmacies). A research project in 2016 aimed at developing a standard procedure for determining the locational potential of pharmacy facilities in Austria according to the Pharmacy Act. The focus is on the estimation of the customer potential of (existing) pharmacies by integrating different data layers in order to create a small-scale GIS model. Therefore, a multiple regression model was developed, which includes and weights the turnover contributions from various demand factors and consumers. An essential element of the analysis is an interaction model, which relates to the attratctiveness of pharmacy facilities and thus a distance-dependent component of the demand-side factors and the utilization of pharmacies by consumers in a spatial perspective. This analogy to a gravitational model prepares the path from previous discrete considerations to a more realistic probability model.

This paper will examine the challenges of combining different data sources and focuses on the triangulation of statistical and GIS modelling. Furthermore, ways of dealing with different spatial resolution of data and quality assurance measures will be discussed by using full data sets for the study on the locational potential of pharmacies in Austria.

Gloria Cicerone (p), Vittorio Carlei, Viktor Venhorst

In this paper we look at a new approach based on co-occurrence analysis in order to assess the skill-relatedness between industries represented in a minimum spanning tree network. The Industry Space, that is the resulting network that connects industries with proximities in terms of skill requirements, is proven to be a valuable tool to lead the opportunities for alternative paths of growth and diversification. In other words the Industry Space we defined is useful to test whether and how the local development is biased positively toward industries that are in the same neighborhood of the industries already present.

Beatriz Solana-De-la-Cuesta (p), Maria Jesus Sanchez-Naranjo, Eva Ponce-Cueto, Joaquin Solana-Gutierrez (p)

The collection of data on trade and population is generally carried out through national and regional census offices, however data relating to traffic are obtained from the information provided by the municipalities, currently in real time and through Geographic Information Systems. The supply of markets and customers within large cities such as Barcelona generates serious congestion problems that make the “last mile distribution” difficult and costly.
The different data source and the varied scale in data collection produce a large panoply of downscaling procedures. In the present study carried out in Barcelona, the data obtained by city districts are disaggregated in raster units by a combination of principal component analysis and discriminant analysis. As a result, raster imputation produced a best imputation with only 11% of information losses versus 30% of leak of information produced by classical imputation by quartiers.
The collection of data on trade and population is generally carried out through national and regional census offices, however data relating to traffic are obtained from the information provided by the municipalities, currently in real time and through Geographic Information Systems. The supply of markets and customers within large cities such as Barcelona generates serious congestion problems that make the “last mile distribution” difficult and costly.
The different data source and the varied scale in data collection produce a large panoply of downscaling procedures. In the present study carried out in Barcelona, the data obtained by city districts are disaggregated in raster units by a combination of principal component analysis and discriminant analysis. As a result, raster imputation produced a best imputation with only 11% of information losses versus 30% of leak of information produced by classical imputation by quartiers.