Papers by Giuseppe Giordano
In this paper we propose to consider the philosophy of the Response Surface Methodology as a furt... more In this paper we propose to consider the philosophy of the Response Surface Methodology as a further graphical resource in reading and interpreting the results of the Correspondence Analysis. We suggest to combine in a threedimensional plot a response surface as a function of different pairs of the more meaningful factorial axes which act as independent ariables. The quantitative response variables to be considered in building the surface can be derived from ifferent sources. We will distinguish an Internal and an External analysis. In the first case, the surface represents some information about the quality of the analysis results. Each point can be properly evaluated according to homogenous regions (isoquant curves) on the factorial plan. In the external analysis we deal with continuous illustrative variables. The proposed explorative tool allows to explore the surface shape as a function of the principal axes of Correspondence Analysis. The Response Surface applied in the framewo...
Applied Stochastic Models and Data Analysis, 2017
The three-way model has been proposed as a development of the original Lee- Carter (LC) model whe... more The three-way model has been proposed as a development of the original Lee- Carter (LC) model when a three-mode data structure is available. The three-way LC model allows enriching the basic LC model by introducing several tools of exploratory data analysis. Such exploratory tools allow giving a new perspective to the demographic analysis supporting the analytical results with a geometrical interpretation and a graphical representation. From a methodological point of view, there are several issues to deal with when focusing on such kind of data. Specially, in presence of the three-way data structure, several choices on data pre-treatment could affect the whole data modelling. The first step of a three-way mortality data investigation consists in exploring the different source of variations and highlighting the significant ones. We will consider the three-way LC model investigated through a three-way analysis of variance with fixed effects, where each cell is given by the mortality rate in a given year of a specific age-group for a country. Firstly, we consider the variability attached to each of the three ways main effects: age, years and countries. Then, we consider the variability induced by the interactions between each pair of the three ways. Finally, the three-way interaction could give information on which country have a specific trend (along years) in each age-group. This kind of analysis is useful to assess the source of variation in the raw mortality data, before to extract rank-one components by the LC-model
Demography and Health Issues, 2018
The costs of the social security public systems, in almost all developed countries, are affected ... more The costs of the social security public systems, in almost all developed countries, are affected by two phenomena: an increasing survival in higher ages and a smaller number of births. The combination of these two aspects largely impacts on the societies dealing with the rising pension and healthcare costs. In spite of the common trend given by the ageing population and the growing longevity, the mortality rates are also influenced by gender, countries, ethnicity, income, wealth, causes of death and so on. According to the WHO a “right” recognition of the causes of death is important for forecasting more accurately mortality. In this framework we intend to investigate the main causes of death impacting on the upcoming human survival, throughout a Multi-dimensional Data Analysis approach to the Lee Carter model of mortality trends. In a previous paper, we stated that the crude mortality data can be considered according to several criteria. In this contribution we take into account a three way array holding mortality data structured by time, age-group and causes of death. The model decomposition we propose is a modified version of the classical Lee Carter model allowing for three-way data treatment, analysis of residuals, graphical representation of the different components. A case study based on actual data will be discussed.
The purpose of this study is to explore how the multimode network approach can be used to analyse... more The purpose of this study is to explore how the multimode network approach can be used to analyse network patterns derived from student mobility flows. We define a tripartite network based on a three-mode data structure, consisting of Italian provinces of residence, universities and fields of study, with student exchanges representing the links between them. A comparison of algorithms for detecting communities from tripartite networks based on modularity optimization is provided, revealing relevant information about the phenomenon under analysis over time. The findings are applied to a real dataset containing micro-level longitudinal information on Italian university students\u2019 careers
I distretti industriali italiani sono entita socio-territoriali caratterizzate da una forte etero... more I distretti industriali italiani sono entita socio-territoriali caratterizzate da una forte eterogeneita, dovuta alla differente composizione, localizzazione, specializzazione produttiva (Core-business) e numerosita delle aziende in esso operanti. Essi risultano essere molto differenziati tra loro anche in base ai modelli di Governance adottati che influenzano la gestione delle relazioni tra gli attori distrettuali e la pianificazione di attivita comuni strumentali allo sviluppo competitivo del distretto. In un quadro cosi complesso si inserisce la nostra ricerca, il cui obiettivo e di descrivere la relazione tra la presenza di organismi e/o strumenti di Governance e i risultati economico-finanziari dei distretti industriali italiani. Il nostro studio si articola in due differenti fasi di analisi. La prima fase - Analisi Qualitativa - riporta una dettagliata descrizione dei distretti industriali italiani, condotta al fine di ottenere un’approfondita conoscenza delle loro dinamiche e...
Social Indicators Research, 2020
The spread of Internet and online social media has created a huge amount of data able to provide ... more The spread of Internet and online social media has created a huge amount of data able to provide new insights to researchers in different disciplinary fields, but it also presents new challenges for data science. Data arising from online social networks can be naturally coded as relational data in affiliation and adjacency matrices, then analyzed with social network analysis. In this study, we apply an interdisciplinary approach (based on automatic visual content analysis, social network analysis, and exploratory statistical techniques) to define and derive a suitable indicator for characterizing places, along with the online activities of travelers, in terms of sharing images. We envisage a novel storytelling perspective where stories are related to places and the narrative activity is realized through posting images. Specifically, we use data extracted from an online social network (i.e., Instagram) to identify travelers' paths among sites of interests. Starting from a large collection of pictures geolocalized in a pre-specified set of locations (i.e., five locations in the Campania region of Italy during the 2018 Christmas season), we use automatic alternative text to produce an ex-post taxonomy of images on the most recurrent themes emerging from pictures posted on Instagram. Quantitative measures defined on the co-occurrence of locations and the emerging themes are used to build a statistical indicator able to characterize paths among different locations as narrated from travelers' posts. The proposed analysis, presented and discussed along with real data, can be useful for stakeholders interested in the fields of policy-making, communication design, and territory profiling strategies.
The study of the scientific collaboration networks is one of the traditional areas of interest in... more The study of the scientific collaboration networks is one of the traditional areas of interest in Network Analysis framework. The aim of the paper is to explore co-authorship networks, where researchers are connected according to the number of papers published together, in order to assess the attitude to collaborate and to identify peculiar styles of collaboration. Starting from the data base of the published papers in the period 1998-2006 produced by the members of the Department of Economics and Statistics at the University of Salerno, collaboration networks are described both by means of network measures (density, actors centrality and presence of cohesive groups) and exploratory multidimensional techniques.
Social Networks
Abstract Multiplex networks arise when more than one source of relationships exists for a common ... more Abstract Multiplex networks arise when more than one source of relationships exists for a common set of nodes. Many approaches to deal with this kind of complex network data structure are reported in the literature. In this paper, we propose the use of factorial methods to visually explore the complex structure of multiplex networks. Specifically, the adjacency matrices derived from multiplex networks are analyzed using the DISTATIS technique, an extension of multidimensional scaling to three-way data. This technique allows the representation of the different types of relationships in both separate spaces for each layer and a compromise space. The analytical procedure is illustrated using a real world example and simulated data.
Studies in Classification, Data Analysis, and Knowledge Organization, 2014
Starting from the main idea of Symbolic Data Analysis to extend Statistics and Data Mining method... more Starting from the main idea of Symbolic Data Analysis to extend Statistics and Data Mining methods from first-order to second-order objects, we focus on network data-as defined in the framework of Social Network Analysis-to define a graph structure and the underlying network in the context of complex data objects. A Network Symbolic description is defined according to the statistical characterization of the network topological properties. We use suitable network measures, which are represented by means of symbolic variables. Their study through multidimensional data analysis, allows for the synthetic representation of a network as a point onto a metric space. The proposed approach is discussed on the basis of a simulation study considering three classical network growth processes.
In this paper we propose to bring together some facets of the Response Surface Methodology used i... more In this paper we propose to bring together some facets of the Response Surface Methodology used in Design of Experiments and the graphical displays arising from Multidimensional Data Analysis in order to enhance the interpretation of the traditional plots of the Factorial Techniques. By exploiting the peculiar interpretation of a response surface we describe a data-fitting based approach in order to analyse the relationships between an outer quantitative response variable and a set of principal axes. Moreover, we to use this interpretative-aid tool to represent by a surface response the information about the quality of the representation derived by a multidimensional data analysis.
Conjoint Analysis is one of the most widely used techniques in the assessment of the consumer’s b... more Conjoint Analysis is one of the most widely used techniques in the assessment of the consumer’s behaviors. This method allows to estimate the partial utility coefficients according to a statistical model linking the overall note of preference with the attribute levels describing the stimuli. Conjoint analysis results are useful in new-product positioning and market segmentation. In this paper a cluster-based segmentation strategy based on a new metric has been proposed. The introduced distance is based on a convex linear combination of two Euclidean distances em bedding information both on the estimated parameters and on the model fitting. Market segments can be then defined according to the proximity of the part-worth coecients and to the explicative power of the estimated models.
Over the past few decades Social Network Analysis has found increasing application in many social... more Over the past few decades Social Network Analysis has found increasing application in many social research areas to describe relational ties among social entities. In this paper, we propose to use Multidimensional Data Analysis in the framework of SNA in order to explore the structural properties of a network. In particular, we refer to Contiguity Analysis in order to deal with relational data defined by ties among the actors and described by network centrality and clustering coefficients. The expected results consist of the definition of a network meta-analysis able to synthesise and visualise the pattern of social relationships in a metric space where the related data structure is described. The proposed method is applied to an illustrative example in the context of a virtual learning community.
In this paper we deal with the study of Social Networks by look ing at a new characterisation of ... more In this paper we deal with the study of Social Networks by look ing at a new characterisation of the net structure. A network can be repr esented according to its graph structure as the entity G (N, E), whereN is the set of nodes and E is the set of edges. Different statistical measures have been defined in order to describe social networks (i.e. centrality, density, cohesion, etc.) (Wasserman and Faust , 1994). Recently, the statistical modeling of random networks has widely used to describe soci al networks (Koehly and Pattison, 2005). The statistical exponential family model s (Strauss and Ikeda, 1990) are a generalization of the Markov random network models int roduced by Frank and Strauss (1986). These models recognize the complex depende ncies within relational data structures and allow to describe a network as a function of a v ector of parameters and a known vector of graph statistics (see eq.1).
Lecture Notes in Social Networks
The presence of patients affected by different diseases at the same time is becoming a major heal... more The presence of patients affected by different diseases at the same time is becoming a major health and societal issue. In clinical literature, this phenomenon is known as comorbidity, and it can be studied from the administrative databases of general practitioners’ prescriptions based on diagnoses. In this contribution, we propose a two-step strategy for analyzing comorbidity patterns. In the first step, we investigate the prescription data with association rules extracted by a two-mode network (or bipartite graph) to find frequent itemsets that can be used to assist physicians in making diagnoses. In the second step, we derive a one-mode network of the diseases codes with association rules, and we perform the k-core partitioning algorithm to identify the most relevant and connected parts in the network corresponding to the most related pathologies.
Aim of this paper is to present a way to explore time series by combining some fundamental result... more Aim of this paper is to present a way to explore time series by combining some fundamental results of stochastic processes theory with graphical and reduction features of factorial methods. A multiple time series visualization and identification strategy is provided by defining a common structural subspace where different regions related to particular ARMA processes are represented. This subspace becomes the reference map for the exploration of multiple time series. In order to provide a unique identification of the ARMA model, a complementary tool represented by a classification tree is proposed.
Uploads
Papers by Giuseppe Giordano
Our study consists of two phases. The first one (Desk Analysis) is a detailed qualitative de-scription of the Italian industrial districts able to provide an in-depth knowledge of their eco-nomic and organizational dynamics. The second one (Quantitative Analysis) deals with the research question about the link between Governance and Performance.
In this study we refer to the Italian districts identified in the 12th annual edition of the Survey on "Le medie imprese industriali italiane (2002-2011)" by Mediobanca and Unioncamere (2013). Furthermore, we use secondary data provided for each company belonging to the dis-trict extracted from the AIDA database.
The complexity of the definition of the object under study together with the high number of observations has been originally addressed by using the theoretical framework of Symbolic Data Analysis (Bock, Diday, 1999).
The use of specific techniques for Multidimensional Data Analysis for Symbolic Data allows to characterize groups of districts, building typologies and to explore the associations between the structure of the districts, their performance and management systems.
In this work we start from a real world business problem and describe the way we model it from a statistical and network analysis point of view, the algorithms we use and the obtained results.
Businesses need to support marketing loyalty campaigns, based on discount programs, in order to boost the customer engagement. More precisely the business’ goal is to get new customers into the program (i.e. customers that were not taking advantage of past campaigns) and increasing the program usage by customers that already used it.
The approach we propose is to use raw transactional data to develop rules to extract recommendations. The recommendations are ego-centric and based on the idea to study the similarity between customers in terms of purchasing behaviour. Profiling each customer with the items that characterise the purchasing of tied customers then produces the recommendations.
We model the raw transactional table as a bipartite graph and, by means of algebraic projection operators and statistical similarity measures; we propose to measure the distance between customers and grouping them into homogeneous clusters. Then, recommendations are produced and ranked in order to extract, for each customer, the best p products to recommend.
In order to describe and discuss the performance of the proposed approach on real data, an experimental campaign was conducted on a customer sample. The results seem to be far better than a previous campaign with different boosting criteria.
changes. Such measures indicate the eciency of the network as a key measure to consider when deciding which nodes to delete from the net, in order to maximize the impact of the deletion. In any real network, some nodes are more highly connected than others are. The eciency measure proposed by (Bienenstock & Bonacich, 2003) is based on a leave-one-out strategy i.e. it is able to detect the nodes whose single removal has the highest impact on the eciency measure which is based on the sum of the lengths of all the shortest paths between all couples of nodes, whose computation is not a simple task itself. By this eciency measure it is possible to evaluate the vulnerability of the whole network with respect to a single node deletion event, but it gives no information
about what could happen to the network if a set of nodes would be cut as a whole. It is easy to show that nding the set of nodes whose deletion has the highest impact on the network is a NP-complete task. It is well known that topological properties influence network resistance to both random or deliberate attacks (Albert, Jeong, & Barabasi, 2000). For instance, Barabasi, Albert and Jeong (1999) showed that an advantage to scale-free architecture is that scale-free networks are resistant to random failures because a few hubs dominate their topology. However, such networks are vulnerable to deliberate attacks on the hubs. Di erent algorithms have been proposed (Fiedler, 1973; Kernighan & Lin, 1970) to con rm these properties (Strogatz, 2001). However, when dealing with large networks such techniques are a ected by computational complexity (Clauset et Al., 2004; Newman, 2004).
In this work we propose to address the more general issue by using Genetic Algorithms (GA). The problem can be described as to nd the smallest set of nodes whose deletion has the highest impact on the eciency criterion mentioned below. GA are powerful heuristics for attacking NP-hard problems. In a very general way, a GA mimics the principles of natural selection (crossover, mutation, selective pressure) to solve problems.
Because Genetic Algorithm complement other optimization methods to help nding good starting points, we can then use traditional optimization techniques to re ne solution. In our study we use the Bienenstock & Bonacich network robustness measure as a tness function and look for subsets of the original network whose deletion most a ects the network connectivity. We weight this measure by the number of removed nodes in order to avoid to get degenerate solutions (removing all the nodes the eciency of the network would be zero). We carry out a screening study based on simulations of di erent network topologies and show some preliminary results. Application on real-world networks have been performed too.