Search

Many international institutions working in the field of food policy have appealed to digitization as the cornerstone of future food system sustainability strategies. However, some works have cast doubts on an excessive emphasis on digitization benefits, highlighting some risk of negative side effects that would hamper, instead of facilitating, food system transition. By crossing the literature on the risks of digitization with that on power and sustainability, the paper stresses the importance of the concept of power for assessing the effects of digital technologies on food system sustainability. One conclusion is that to ensure that digitization is consistent with sustainability objectives, public intervention should address some critical issues, such as: data sovereignty, increased surveillance and corporate control over farming practices, increased influence of corporate power on state regulatory choices.

Ordinal responses in the form of ratings arise frequently in social sciences, marketing and business applications where preferences, opinions and perceptions play a major role. Main examples concern customers/users' satisfaction analysis where it is common to collect rater's evaluation on a hedonic scale, along with a set of covariates (both categorical and quantitative) that characterize the respondent and/or the item/service. In this framework, the ordinal nature of the response has to be properly taken into account when the interest is in the understanding of different response patterns in terms of subjects' covariates. In this spirit, a model-based tree procedure for ordinal scores is illustrated: its structure is based on a class of mixture models for ordinal rating data that implies a twofold analysis in terms of feeling and uncertainty and effective graphical visualization of results. The flexibility of the chosen modelling framework entails that the splitting criterion can be customized according to the purposes of the study and the available data, without disregarding uncertainty. Thus, the selection of variables yielding to the best partitioning results is driven by fitting measures or classical likelihood and deviance measurements, for instance. The contribution proposes to investigate the features of varying decision rules and thus implicitly addresses the problem of selecting the model-based tree that provides the most adequate and satisfying overview of response profiles. Comparison with alternative model-based approaches is also outlined.

Tree based methods in statistics are gaining a renewed interest in the Big Data era since they entail effective interpretation of results. In this setting, we apply a model-based technique to build trees for ordinal responses relying on a class of mixture models whose characteristic feature is the probabilistic specification of uncertainty. An application to the perception of happiness shows that the integration of tree methods with the chosen modelling boosts cluster analysis of respondents.

This study aims at analysing consumer behaviour towards online advertising, focusing on the attitude defined in terms of cognitive, affective and behavioural aspects. Likert scale was employed to capture these three aspects. However, subjective attitude is often a nebulous and vague feature to deal with. To this end, adopting a fuzzy clustering procedure, first a fuzzy coding of the levels of the Likert scale is adopted. Second, the Fuzzy k-Means clustering model for fuzzy data has been employed to segment a sample of web users. Two clusters have been identified that require different marketing communication strategies.

In this paper we propose a method to locate multiple structural breaks in financial time series that accounts for the interval structure of these series as well as for the presence of outliers. For each time unit, the upper and lower bound of the intervals depend on the closing value. Then, to locate the break dates, a robust exponential based distance, that is able to neutralize the impact of outlier, is employed in the framework of Atheoretical Regression Trees. An empirical application to the prices of an asset shows the usefulness of the proposed procedure.

The detection of change points in time series is a popular subject of research and the most challenging task is to identify multiple changes occurring at unknown date. At this aim [2] proposed a method called ART (Atheoretical Regression Trees) that employs least squares regression trees to detect multiple breaks in the mean. In this paper we present a review of the method as well as two recent extensions meant to deal either with changes in the coefficients of a parametric model or with changes in time series imprecisely observed i.e. time ordered observations whose values are not known exactly, such as interval or ordinal time series. An empirical example concerning a continuous-valued imprecise time series is presented and discussed.

We propose a cluster analysis approach to quantify near-synonymy relations and compare non-parametric and parametric methods. The first approach is model free since it does not assume an underlying model of lexical knowledge but it uncovers the group structure in the set of near-synonyms of a target word by comparing the list of synonyms of the given entry with those of its near-synonyms as contained into available thesauri. Then, in order to validate the results provided by the cluster analysis, a statistical model is introduced for analyzing human judgments of perceived degree of synonymy, also by a relationship with subjects’ characteristics. Empirical evidence for a selected word of Italian is presented and discussed.

The analysis of structural-change models is nowadays a popular subject of research both in econometric and statistical literature. The most challenging task is to identify multiple breaks occurring at unknown dates. In case of multiple shifts in mean Cappelli and Reale (2005) have proposed a method called ART that employs regression trees to estimate the number and location of breaks. In this paper we focus on regime changes due to breaks in the coefficients of a parametric model and we propose an extension of ART that addresses this topic in the general framework of the linear model with multiple structural changes. The proposed approach considers in the tree growing phase the residuals of parametric models fitted to contiguous subseries obtained by splitting the original series whereas tree pruning together with model selection criteria provide the number of breaks. We present simulation results pertaining to the behavior of the proposed approach as well as an empirical application to the time series of the car drivers killed or seriously injured in Great Britain in traffic accidents aiming at evaluating the impact of UK introduction of compulsory seatbelt wearing.

The analysis of structural-change models is nowadays a popular subject of research both in econometric and statistical literature. The most challenging task is to identify multiple breaks occurring at unknown dates. In case of multiple shifts in mean Cappelli et al (2008) have proposed a method based on regression trees. In this paper we propose an extension of this method to address the problem of estimating the break dates and their number in the general framework of the linear model with multiple structural changes. We present simulation results pertaining to the behavior of the proposed approach.

In this paper we propose a procedure to detect the presence of co-breaking i.e. of a common structural break occurring at an unknown date in a vector of time series. Co-breaking occurs if a linear combination of the time series cancels the break. Our procedure employs a regression tree based approach called ART to detect the presence of breaks and Principal Component Analysis to generate the linear combinations of the vector series. On each of these linear combinations ART is performed again to detect the presence of a break. The combination that "hides" the co-breaking time is the one minimizing the employed splitting criterion. The results of a simulation study carried out to evaluate the performance of the proposed approach are presented and discussed.

In this paper we propose a procedure to detect the presence of co-breaking i.e. of a common structural break occurring at an unknown date in a vector of time series. Co-breaking occurs if a linear combination of the time series cancels the break. Our procedure employs a regression tree based approach namely ART, to detect the presence of breaks in the mean of each one-dimensional component of a multidimensional time series and Principal Component Analysis to estimate the linear combinations of the vector series. On each of these linear combinations ART is performed again to detect the presence of a break. The combination that "hides" the co-breaking time is the one minimizing the employed splitting criterion. The results of a simulation study carried out to evaluate the performance of the proposed approach are presented and discussed.

In the last two decades a large amount of research both in the econometric and statistics literature has been concerned with the detection of structural changes in time series. The most challenging task is to identify multiple breaks occurring at unknown date and most contributions have addressed the case of level shifts. In this context Cappelli et al (2005, 2008) have proposed a computational efficient procedure called ART that employs regression trees to identify the breaks in the mean and their locations. In this streamline this paper focuses on a different problem: the detection of regime changes due to instability in model parameters. At this aim we propose an extension of the ART procedure that uses in the tree growing stage the residuals of models fitted to contiguous subseries obtained by splitting the original series. The best split is the one that maximizes the reduction in the residuals when splitting a node into its offsprings; the location of the splits provides the dates at which the regime change occurred. For the purpose to select of the final number of breaks the Chow test (1960) can be employed: splitting stops if the achieved reduction does not ensure the chosen significance level. Eventually, in order to circumvent the problem of model mispecification, in our programme various models can be assumed growing candidate trees (i.e., sets of breaks). The performance of the proposed approach is evaluated by means of a simulation study. An application to the US labor productivity index is also presented and discussed

In the last decade lot of interest, both in the econometric and statistical literature, has been paid to the issue of confusing long memory and structural breaks in level. Indeed, there is evidence that a stationary short memory process that encounters occasional level shifts can show a slow rate of decay in the autocorrelation function and other properties of I(d) processes, where d can be a fraction. On the other hand, when the data generating process is an integrated or fractionally integrated process (with no breaks), several breaks can be detected spuriously. Thus, testing procedures that allow a distinction between long memory and level shifts are currently a popular subject of research. Here we show the results of a simulation study on a simple empirical strategy that can be applied as a first check, to assess "what is what". This strategy consists in fitting long memory and structural break models separately; in case both provide plausible explanation of the DGP of the data at hand, the long memory and structural break analysis are repeated on the series break-free and on the filtered series, respectively. Then, the presence of breaks in the filtered series and/or of long memory behavior in the break-free series indicate whether the series shows real or spurious breaks and long memory.

In this paper we propose a strategy aimed at variable selection in models for ordinal data. The procedure core exploits the primary splitters, as they arise from a tree-based procedure. Indeed, the latter is based on a recursive partitioning approach that increasingly reduces the so called impurity of the response variable based on the covariate values. Since the reduction in impurity provides a natural importance ranking of the covariates, this allows to identify those relevant among the available ones. The proposed strategy is, then, used for detecting relevant covariates in a case study concerning the consumers’ preferences towards different types of smoked salmons.

ART was introduced as a modified use of CART methodology for quick detection of structural breaks in the mean levels. In this paper simulations are presented to test ART against a number of different types of time series, to find a good pruning method, and to compare with alternative approaches.

In this paper we propose a computationally effective approach to detect multiple structural breaks in the mean occurring at unknown dates. We propose a non-parametric approach that exploits, in the framework of least squares regression trees, the contiguity property of the Fisher grouping method (1958) proposed for grouping a single real variable. The proposed approach is applied to study the possibility of using the series of anomalous observation C17 provided by the seasonal adjustment procedure implemented in X12-ARIMA.

In this paper we propose a non parametric approach for detecting multiple structural break occurring at unknown dates. The method exploits, in the framework of least squares regression trees, the contiguity property introduced by Fisher for grouping a single real variable. The proposed approach called ART is applied to the structural breaks analysis of the U.S. Ex-Post Real Interest Rate series.

This paper proposes a non parametric approach for dating structural breaks whose number and dates of occurrence are a priori unknown. In particular, the case of level shifts is considered. For the purpose of locating the break-dates the method exploits, in the framework of least square regression trees, the contiguity property introduced by Fisher for grouping a single real variable. The proposal is applied to the study of the mean water level of Michigan Huron lake also comparing the results to those of the well known procedure recently proposed by Bai and Perron.

Nell’ambito dei metodi di classificazione e regressione ad albero si mostrerà come l’algoritmo STP (Cappelli et al, 2001) costituisca un utile strumento di overfitting avoidance. Il problema dell’ overfitting si sostanzia nella presenza di “false” suddivisioni che riflettono peculiarità dei dati piuttosto che relazioni realmente esistenti tra la variabile di risposta ed i predittori. I metodi classici utilizzati per far fronte a tale problema essendo basati sulla valutazione dell’overfitting in termini di errore appaiono non del tutto appropriati all’uopo. Per contro, la procedura STP, essendo impostata sulla valutazione della relazione di dipendenza tra la variabile di risposta e le variabili di split, è in grado di valutare la presenza di sovradattamento e conservare le sole suddivisoni realmente esplicative.

Nel presente lavoro viene proposto un approccio per la individuazione di break strutturali in serie temporali quando si dispone di serie esplicative. Il metodo si articola in uno schema iterativo basato sull'uso alternato della analisi canonica e della regressione ad albero. In particolare la informazione contenuta nelle serie esplicative viene sintetizzata mezzo della analisi canonica in combinazioni lineari successivamente usate come covariate nella regressione ad albero. la proposta viene illustrata mediante uno studio di simulazione

In this paper the problem of identifying the structure of expansions and recessions in the US economy is placed in the framework of recursive partitioning and discriminant analysis. The classification provided by the National Bureau of Economic Research (NBER) is considered. Using as covariates the main variables and indicators deemed useful to predict the business cycle, a modified recursive partitioning approach is proposed; at each step (tree node) the method identifies the linear combination of the covariates that discriminates the most between being in and out of a recession; this new covariate is then used to split the data. The application to the case of the US business cycle and the comparison with classical logistic regression shows the merits of the proposed approach that represents a useful to tool to identify and to interpret the structure of expansions and recessions.

The focus of this paper is on synonymy relations between words. A cluster analysis approach is presented , aiming at partitioning the semantic space of a word -suitably defined- in regions (i.e. subgroups of synonyms of the target word) charcterised by a high degree of homogeneity and therefore interchangeability. Application to the case of italian words are presentde and discussed showing that the proposed method is able to identify different senses of a word.

his paper deals with the problem of dimension reduction in the general context of supervised statistical learning, with particular attention to data mining applications. The main goal of the proposed methodology is to improve tree based methods as prediction tool by introducing an alternative approach to data partitioning which is meant to handle large numbers of (possibly correlated) covariates. The key idea is to use suitable combinations of covariates recursively identified.

This paper provides a maximum likelihood estimation strategy to identify a tree-based model which, being a function of a set of observed optimally pruned trees, represents the nal classication model. The strategy is based on a probability distribution and it uses a metric based on structural dierences among trees. An example on a real dataset is also presented to show how the procedure works.

This paper provides a new pruning method for classification trees based on the impurity-complexity measure. Advantages of the proposed approach compared to the error-complexity pruning method are outlined showing an example on a real data set.

We propose a recursive partitioning approach to identify groups of risky financial institutions using a synthetic indicator built on the information arising from a sample of pooled systemic risk measures. The composition and amplitude of the risky groups change over time, emphasizing the periods of high systemic risk stress. We also calculate the probability that a financial institution can change risk group over the next month and show that a firm belonging to the lowest or highest risk group has in general a high probability to remain in that group.

Ordinal responses in the form of ratings arise frequently in social sciences, marketing and business applications where preferences, opinions and perceptions play a major role and it is common to collect along with rater’s evaluation, a set of covariates that characterize the respondent and/or the item/service. In this framework, the ordinal nature of the response has to be properly taken into account when the interest is in the understanding of different response patterns in terms of subjects’ covariates. In this spirit, a model-based tree procedure for ordinal scores is illustrated: its structure is based on a class of mixture models for ordinal rating data that implies a twofold analysis in terms of feeling and uncertainty and effective graphi- cal visualization of results. The chosen modelling framework entails that the splitting criterion can be customized according to the purposes of the study and the available data. Thus, the selection of variables yielding to the best partitioning results is driven by fitting measures or classical likelihood and deviance measurements, for instance. In order to illustrate the performances of the different splitting criterion for the cubremot procedure, we consider data from the 5th European Working Condition Survey carried out by Eu- rofound in 2010, comparison with alternative approaches that grow trees for ordinal responses is also outlined.

In this paper a novel approach is proposed for ranking data analysis that is based on marginal distributions. The marginal scores for one of the items to be ranked can be considered as a proxy of preference ratings and thus they can be employed to investigate drivers of the outcome. The rationale is that a model for the univariate marginal distributions should suitably convey information about the veracity of ordinal scores given that it would disregard the ranking constraints. The method relies on cubremot ( cub re gression mo del t rees), a tool for growing trees for ordinal resposes based on the local estimation of cub models, a class of models able to explain both preferences and related uncertainty. Here we propose the Uncertainty Tree to disclose preference patterns on the basis of the model uncertainty component. Two applications on real data are discussed; comparison with other model-based trees is provided and a synthetic index for the selection of the best cubremot is also advanced.

The paper introduces a new technique for growing trees for ordinal responses in the model-based framework. We consider the class of cub mixtures which is particularly appropriate to model perceptions and evaluations, as it designs the response process as the combination of a personal feeling and an inherent uncertainty. In our proposal, the partitioning process is based on the local estimation of cub regression models to profile responses according to feeling and uncertainty conditional to the splitting variables. In this regard, two alternative splitting criteria are introduced featuring both inferential and fitting issues. Moreover, the chosen modelling framework allows for advantageous visualization of the classification results. The proposal is illustrated using real data from a survey conducted by the Italian National Statistical Office, with focus on perceived trust towards the European Parliament. A benchmark study is conducted to settle the proposal among the available tree methods for ordinal responses.

CUBREMOT (CUB REgression MOdel Trees) is a model-based approach to grow trees for ordinal responses that relies on a class of mixture models for evaluations and preferences (CUB). The original proposal considers deviances in loglikelihood to partition observations. In the present paper a new splitting criterion is introduced that, among the significant splitting variables, chooses the one that maximizes a dissimilarity measure. This choice is tailored to generating child nodes as far apart as possible with respect to the estimated probability distributions. An application to real data on Italians’ trust towards the European Parliament taken from the official survey on daily life conducted by the Italian National Institute of Statistics (ISTAT) in 2015 is presented and discussed in comparison with alternative methods.

The focus of the contribution is on the splitting criterion for model-based treeprocedure based on the class ofCUBmixture models for the analysis of ordinal scores. Theflexibility of the chosen modelling framework allows to select the splitting criterion to growthe tree according to the purposes of the study and the available data. In particular, theselection of variables yielding to the best partitioning results can be driven by fitting measuresor classical likelihood and deviance measures. The contribution proposes to investigate thefeatures of the available decision rules by a set of Montecarlo experiments, thus implicitlyfacing the problem of selecting the model-based tree to obtain an adequate and satisfyingoverview of response profiles.