Rémy Le Boennec (p), Fouad Hadj Selem

In the Ile-de-France Région, France, the modal share of car in commuting trips is 42% (Source: Global Transport Survey, 2010), a much lower figure than the national average (70%). In this article, we use an original predictive tool that relies on multi-source input data to estimate the flows, modes, and socio-demographic characteristics of mobility users. We focus on commuting, as the recurrence of such trips theoretically enhances modal shift towards more sustainable transport modes on the long term.

From the output data predicted by the tool, we elaborate, for all origin-destination couples (O-D) between municipalities in the Ile-de-France Région (n = 1286²), an original indicator of the relative performance of public transport in relation to car. We measure this indicator as compared travel times: for each O-D couple, we compare the predicted travel time to the modal share of car, also predicted by our tool. We thus identify three groups of municipalities:

- A first group whose respective modal shares of car and public transport are consistent with the modal performance indicator: comparable travel times between car and public transport are correlated with respective modal shares close to the Ile-de-France average;
- A second group whose residents seem more favorable to public transport than suggested by the modal performance indicator, as predicted travel times appear to be in favor of car;
- A third one, with the opposite conclusions.

A more detailed territorial analysis focuses, in a second time, to cross these three groups of municipalities with the following typology of territories: Paris (central city), other municipalities of Greater Paris (inner suburbs), other municipalities of the Paris agglomeration (outer suburbs), central municipalities of the rest of the Ile-de-France Région, other municipalities. This typology crosses dwelling densities, employment densities and flows between the two, on the one hand, and a morphological criterion delimiting the boundaries of the agglomeration, on the other hand. The interest of such a crossover lies in mobility issues that remain specific, within the Ile-de-France, to each type of area.

Such results highlighting modal preferences of households appear valuable for social sciences, in that they allow to empirically raise the calibration difficulties often pointed at in theoretical models of urban economics.

María Jesús Delgado Rodríguez (p), César Pérez López (p), Sonia de Lucas Santos (p)

The goal of the present research is to contribute to detection of tax fraud concerning personal income tax returns (IRPF, in Spanish) filed in Spain, through the use of Machine Learning advanced predictive tools, by applying Multilayer Perceptron neural network (MLP) models. The possibilities springing from these techniques have been applied to a broad range of personal Income return data supplied by the Institute of Fiscal Studies (in Spanish, Instituto de Estudios Fiscales IEF). The use of the networks enabled taxpayer segmentation as well as calculation of the probability concerning an individual taxpayer’s propensity to attempt to evade tax. The results showed that the selected model has an efficiency rate of 84.3%, implying improvement in relation to other models utilized in tax fraud detection. The proposal can be generalised to quantify an individual’s propensity to commit fraud as regards other kinds of taxes. These models will support tax offices to arrive at the best decisions regarding action plans to combat tax fraud.

Francisco Rowe (p), Juan Soto

Administrative areas do not accurately reflect the contemporary spatial manifestation of labour market linkages. Local labour market areas (LLMAs) have been shown to provide a better representation of geographic labour market activity. Traditionally LLMAs are delineated based on commuting flow data. However, communting data are expensive to collect, are sporadically collected in developed countries and rarely available in less developed countries. Yet, recent advances in computing capacity and increased availability of satellite imagery offers a unique opportunity to generate LLMAs in poor environment context in a cheap, frequent and automated way. This papers aims to develop an automated satellite-based approach to define LLMAs.

Pavlos Baltas (p), Demetris Stathakis , Leonidas Liakos

Digital traces are all we leave while surfing on the internet, especially when we use social media. Every regional policy proposal and/or assessment relies on sound and timely data. Can we use these digital traces in order to produce reliable data series? Here we explore the use of Big Data for estimating seasonal population variations. Timely monthly population estimates are important in regional science, especially when we are studying phenomena such as tourism, where seasonality is a key aspect. The objective of this work is to produce timely and informative monthly estimates of seasonal population variations based on social media data. The proposed method is based in a new source, the Facebook’s Advertising Platform, which is also open-domain. The method is applied in Greek Island regions. The majority of Greek islands have a tourism-depended economy. Also, due to tourism, population size changes month by month and makes the selected regions the ideal laboratory to test the method. The results of the proposed method are compared with these obtained by conventional sources as arrivals by air and sea, night time satellite images. Monitoring and controlling seasonal population pressures in islands is a key factor due to limited environmental problems. For example, water use in regions that face water scarcity (like islands in Mediterranean) is increased due to seasonal variation of population’s concentration in time and space. Population estimates by month can play an important role as a means of measuring pressure and identifying environmental limits (carrying capacity) of a region.