D4.3 - Computational analysis report

This document reports the results of the different computational analyses performed on the data collected in the project. After a brief introduction, the document includes several sections devoted to different kinds of analysis. In each section, we have first an introduction on the objectives of the analysis presented in the context of the Diamond project, then it offers a description of the methodology, and the presentation of the results obtained, organized by use case. Finally, a last section is devoted to discussion and general conclusions. The computational methods employed for the data analysis include: Analysis of structured open data to define indicators that characterize the area surrounding each urban railway station (use case I) or docking station (Use case III) along socio-demographic, territory and mobility aspects, and to select relevant stations for the subsequent steps of the analysis (Section 2). Analysis of observations, to characterize the selected urban railways infrastructures (Use case I) and docking stations (Use case III) (Section 3). Analysis of experimental sessions in a simulator laboratory setting, to validate the impact of autonomous vehicles driving experience and passengers’ emotional appraisal, considering gender and other relevant intersectional variables (Use case II), combined with the analysis of user satisfaction surveys for Use case II (Section 4). Analysis of social media data collected from Twitter, to get insights on the concerns of men and women regarding topics related to Use case I, II and III (Section 5) Analysis of user satisfaction surveys on service provisions (Use cases I and III) and employment conditions (Use case IV), exploring and assessing differences in user satisfaction according to their socio-demographic characteristics, by means of statistical tests, regression analysis, factor analysis (Section 6). Analysis of users and employees' priorities from Dynamic Augmentative Delphi surveys, employing Analytic Hierarchy Processes to assess the ranking between priorities for different groups of respondents (Section 7). Bayesian network analysis to obtain a weighted hierarchy of fairness characteristics, based on open and proprietary structured data, observations and UESI questionnaires, and a statistical assessment of significant differences between users along socio-demographic variables (Section 8). While the document is structured along the different kinds of analysis performed, the final conclusions present a summary of the main results, organized per use case (Section 9).