arXiv subjects

Giuseppe De Nicolao

Publications and source records attributed to Giuseppe De Nicolao.

17 recordsLinked to original sources

arXiv2026

An Interval-Score ROC Curve for Assessment, Calibration and Ensembling of Probabilistic Forecasts

Probabilistic forecast evaluation is inherently multi-objective, yet existing proper scoring rules reduce predictive performance to a single scalar value, potentially obscuring the trade-off between forecast concentration and predictive accuracy. We introduce the Interval-Score Receiver Operating Characteristic (IS-ROC) Curve, a graphical framework that represents the complete family of interval forecasts generated by varying prediction tightness. We show that the IS-ROC Curve induced by the data generating process is Pareto optimal and convex, providing a geometric characterization of the optimal forecasting frontier. Building on these properties, we propose a geometry-based calibration procedure based on tangent optimization and convexification, together with an ensemble strategy that combines competing forecasters through convex hull construction. Finally, we provide a practical workflow and numerical examples illustrating forecast comparison, calibration, and ensemble construction within the proposed framework.

stat.ME↗

arXiv2023

Correction of Italian under-reporting in the first COVID-19 wave via age-specific deconvolution of hospital admissions

When the COVID-19 pandemic first emerged in early 2020, healthcare and bureaucratic systems worldwide were caught off guard and largely unprepared to deal with the scale and severity of the outbreak. In Italy, this led to a severe underreporting of infections during the first wave of the spread. The lack of accurate data is critical as it hampers the retrospective assessment of nonpharmacological interventions, the comparison with the following waves, and the estimation and validation of epidemiological models. In particular, during the first wave, reported cases of new infections were strikingly low if compared with their effects in terms of deaths, hospitalizations and intensive care admissions. In this paper, we observe that the hospital admissions during the second wave were very well explained by the convolution of the reported daily infections with an exponential kernel. By formulating the estimation of the actual infections during the first wave as an inverse problem, its solution by a regularization approach is proposed and validated. In this way, it was possible to computed corrected time series of daily infections for each age class. The new estimates are consistent with the serological survey published in June 2020 by the National Institute of Statistics (ISTAT) and can be used to speculate on the total number of infections occurring in Italy during 2020, which appears to be about double the number officially recorded.

stat.AP↗

arXiv2021

Regularization methods for the short-term forecasting of the Italian electric load

The problem of forecasting the whole 24 profile of the Italian electric load is addressed as a multitask learning problem, whose complexity is kept under control via alternative regularization methods. In view of the quarter-hourly samplings, 96 predictors are used, each of which linearly depends on 96 regressors. The 96x96 matrix weights form a 96x96 matrix, that can be seen and displayed as a surface sampled on a square domain. Different regularization and sparsity approaches to reduce the degrees of freedom of the surface were explored, comparing the obtained forecasts with those of the Italian Transmission System Operator Terna. Besides outperforming Terna in terms of quarter-hourly mean absolute percentage error and mean absolute error, the prediction residuals turned out to be weakly correlated with Terna, which suggests that further improvement could ensue from forecasts aggregation. In fact, the aggregated forecasts yielded further relevant drops in terms of quarter-hourly and daily mean absolute percentage error, mean absolute error and root mean square error (up to 30%) over the three test years considered.

eess.SP↗

arXiv2021

Identification of AC Networks via Online Learning

The increasing penetration of intermittent distributed energy resources in power networks calls for novel planning and control methodologies which hinge on detailed knowledge of the grid. However, reliable information concerning the system topology and parameters may be missing or outdated for temporally varying electric distribution networks. This paper proposes an online learning procedure to estimate the network admittance matrix capturing topological information and line parameters. We start off by providing a recursive identification algorithm exploiting phasor measurements of voltages and currents. With the goal of accelerating convergence, we subsequently complement our base algorithm with a design-of-experiment procedure which maximizes the information content of data at each step by computing optimal voltage excitations. Our approach improves on existing techniques, and its effectiveness is substantiated by numerical studies on realistic testbeds.

eess.SY↗

arXiv2021

Preemptive periodic epidemic control reduces life and healthcare system costs without aggravation of social and economic losses

Many countries are managing COVID-19 epidemic by switching between lighter and heavier restrictions. While an open-close and a close-open cycle have comparable socio-economic costs, the former leads to a much heavier burden in terms of deaths and pressure on the healthcare system. An empirical demonstration of the toll ensuing from procrastination was recently observed in Israel, where both cycles were enforced from late August to mid-December 2020, yielding some 1,600 deaths with open-close compared to 440 with close-open.

eess.SY↗

arXiv2021

Vaccination and SARS-CoV-2 variants: how much containment is still needed? A quantitative assessment

Despite the progress in medical care, combined population-wide interventions (such as physical distancing, testing and contact tracing) are still crucial to manage the SARS-CoV-2 pandemic, aggravated by the emergence of new highly transmissible variants. We combine the compartmental SIDARTHE model, predicting the course of COVID-19 infections, with a new data-based model that projects new cases onto casualties and healthcare system costs. Based on the Italian case study, we outline several scenarios: mass vaccination campaigns with different paces, different transmission rates due to new variants, and different enforced countermeasures, including the alternation of opening and closure phases. Our results demonstrate that non-pharmaceutical interventions (NPIs) have a higher impact on the epidemic evolution than vaccination, which advocates for the need to keep containment measures in place throughout the vaccination campaign. We also show that, if intermittent open-close strategies are adopted, deaths and healthcare system costs can be drastically reduced, without any aggravation of socioeconomic losses, as long as one has the foresight to start with a closing phase rather than an opening one.

q-bio.PE↗

arXiv2021

Ensembling methods for countrywide short term forecasting of gas demand

Gas demand is made of three components: Residential, Industrial, and Thermoelectric Gas Demand. Herein, the one-day-ahead prediction of each component is studied, using Italian data as a case study. Statistical properties and relationships with temperature are discussed, as a preliminary step for an effective feature selection. Nine "base forecasters" are implemented and compared: Ridge Regression, Gaussian Processes, Nearest Neighbours, Artificial Neural Networks, Torus Model, LASSO, Elastic Net, Random Forest, and Support Vector Regression (SVR). Based on them, four ensemble predictors are crafted: simple average, weighted average, subset average, and SVR aggregation. We found that ensemble predictors perform consistently better than base ones. Moreover, our models outperformed Transmission System Operator (TSO) predictions in a two-year out-of-sample validation. Such results suggest that combining predictors may lead to significant performance improvements in gas demand forecasting.

cs.LG↗

arXiv2021

Forecasting residential gas demand: machine learning approaches and seasonal role of temperature forecasts

Gas demand forecasting is a critical task for energy providers as it impacts on pipe reservation and stock planning. In this paper, the one-day-ahead forecasting of residential gas demand at country level is investigated by implementing and comparing five models: Ridge Regression, Gaussian Process (GP), k-Nearest Neighbour, Artificial Neural Network (ANN), and Torus Model. Italian demand data from 2007 to 2017 are used for training and testing the proposed algorithms. The choice of the relevant covariates and the most significant aspects of the pre-processing and feature extraction steps are discussed in-depth, lending particular attention to the role of one-day-ahead temperature forecasts. Our best model, in terms of Root Mean Squared Error (RMSE), is the ANN, closely followed by the GP. If the Mean Absolute Error (MAE) is taken as an error measure, the GP becomes the best model, although by a narrow margin. A main novel contribution is the development of a model describing the propagation of temperature errors to gas forecasting errors that is successfully validated on experimental data. Being able to predict the quantitative impact of temperature forecasts on gas forecasts could be useful in order to assess potential improvement margins associated with more sophisticated weather forecasts. On the Italian data, it is shown that temperature forecast errors account for some 18% of the mean squared error of gas demand forecasts provided by ANN.

cs.CY↗

arXiv2020

Fast calibration of two-factor models for energy option pricing

Energy companies need efficient procedures to perform market calibration of stochastic models for commodities. If the Black framework is chosen for option pricing, the bottleneck of the market calibration is the computation of the variance of the asset. Energy commodities are commonly represented by multi-factor linear models, whose variance obeys a matrix Lyapunov differential equation. In this paper, analytical and numerical methods to derive the variance are discussed: the Lyapunov approach is shown to be more straightforward than ad-hoc derivations found in the literature and can be readily extended to higher-dimensional models. A case study is presented, where the variance of a two-factor mean-reverting model is embedded into the Black formulae and the model parameters are calibrated against listed options. The analytical and numerical method are compared, showing that the former makes the calibration 14 times faster. A Python implementation of the proposed methods is available as open-source software on GitHub.

q-fin.PR↗

arXiv2020

Stable spline identification of linear systems under missing data

A different route to identification of time-invariant linear systems has been recently proposed which does not require committing to a specific parametric model structure. Impulse responses are described in a nonparametric Bayesian framework as zero-mean Gaussian processes. Their covariances are given by the so-called stable spline kernels encoding information on regularity and BIBO stability. In this paper, we demonstrate that these kernels also lead to a new family of radial basis functions kernels suitable to model system components subject to disturbances given by filtered white noise. This novel class, in cooperation with the stable spline kernels, paves the way to a new approach to solve missing data problems in both discrete and continuous-time settings. Numerical experiments show that the new technique may return models more predictive than those obtained by standard parametric Prediction Error Methods, also when these latter exploit the full data set.

eess.SY↗

arXiv2020

On the agreement between bibliometrics and peer review: evidence from the Italian research assessment exercises

This paper appraises the concordance between bibliometrics and peer review, by drawing evidence from the data of two experiments realized by the Italian governmental agency for research evaluation. The experiments were performed for validating the dual system of evaluation, consisting in the interchangeable use of bibliometyrics and peer review, adopted by the agency in the research assessment exercises. The two experiments were based on stratified random samples of journal articles. Each article was scored by bibliometrics and by peer review. The degree of concordance between the two evaluations is then computed. The correct setting of the experiments is defined by developing the design-based estimation of the Cohen's kappa coefficient and some testing procedures for assessing the homogeneity of missing proportions between strata. The results of both experiments show that for each research areas of hard sciences, engineering and life sciences, the degree of agreement between bibliometrics and peer review is -- at most -- weak at an individual article level. Thus, the outcome of the experiments does not validate the use of the dual system of evaluation in the Italian research assessments. More in general, the very weak concordance indicates that metrics should not replace peer review at the level of individual article. Hence, the use of the dual system of evaluation for reducing costs might introduce unknown biases in a research assessment exercise.

stat.AP↗

arXiv2019

Citation gaming induced by bibliometric evaluation: a country-level comparative analysis

It is several years since national research evaluation systems around the globe started making use of quantitative indicators to measure the performance of researchers. Nevertheless, the effects on these systems on the behavior of the evaluated researchers are still largely unknown. We attempt to shed light on this topic by investigating how Italian researchers reacted to the introduction in 2011 of national regulations in which key passages of professional careers are governed by bibliometric indicators. A new inwardness measure, able to gauge the degree of scientific self-referentiality of a country, is defined as the proportion of citations coming from the country itself compared to the total number of citations gathered by the country. Compared to the trends of the other G10 countries in the period 2000-2016, Italy's inwardness shows a net increase after the introduction of the new evaluation rules. Indeed, globally and also for a large majority of the research fields, Italy became the European country with the highest inwardness. Possible explanations are proposed and discussed, concluding that the observed trends are strongly suggestive of a generalized strategic use of citations, both in the form of author self-citations and of citation clubs. We argue that the Italian case offers crucial insights on the constitutive effects of evaluation systems. As such, it could become a paradigmatic case in the debate about the use of indicators in science-policy contexts.

cs.DL↗

arXiv2017

Errors and secret data in the Italian research assessment exercise. A comment to a reply

Italy adopted a performance-based system for funding universities that is centered on the results of a national research assessment exercise, realized by a governmental agency (ANVUR). ANVUR evaluated papers by using 'a dual system of evaluation', that is by informed peer review or by bibliometrics. In view of validating that system, ANVUR performed an experiment for estimating the agreement between informed review and bibliometrics. Ancaiani et al. (2015) presents the main results of the experiment. Baccini and De Nicolao (2017) documented in a letter, among other critical issues, that the statistical analysis was not realized on a random sample of articles. A reply to the letter has been published by Research Evaluation (Benedetto et al. 2017). This note highlights that in the reply there are (1) errors in data, (2) problems with 'representativeness' of the sample, (3) unverifiable claims about weights used for calculating kappas, (4) undisclosed averaging procedures; (5) a statement about 'same protocol in all areas' contradicted by official reports. Last but not least: the data used by the authors continue to be undisclosed. A general warning concludes: many recently published papers use data originating from Italian research assessment exercise. These data are not accessible to the scientific community and consequently these papers are not reproducible. They can be hardly considered as containing sound evidence at least until authors or ANVUR disclose the data necessary for replication.

cs.DL↗

arXiv2016

Do they agree? Bibliometric evaluation vs informed peer review in the Italian research assessment exercise

During the Italian research assessment exercise, the national agency ANVUR performed an experiment to assess agreement between grades attributed to journal articles by informed peer review (IR) and by bibliometrics. A sample of articles was evaluated by using both methods and agreement was analyzed by weighted Cohen's kappas. ANVUR presented results as indicating an overall 'good' or 'more than adequate' agreement. This paper re-examines the experiment results according to the available statistical guidelines for interpreting kappa values, by showing that the degree of agreement, always in the range 0.09-0.42 has to be interpreted, for all research fields, as unacceptable, poor or, in a few cases, as, at most, fair. The only notable exception, confirmed also by a statistical meta-analysis, was a moderate agreement for economics and statistics (Area 13) and its sub-fields. We show that the experiment protocol adopted in Area 13 was substantially modified with respect to all the other research fields, to the point that results for economics and statistics have to be considered as fatally flawed. The evidence of a poor agreement supports the conclusion that IR and bibliometrics do not produce similar results, and that the adoption of both methods in the Italian research assessment possibly introduced systematic and unknown biases in its final results. The conclusion reached by ANVUR must be reversed: the available evidence does not justify at all the joint use of IR and bibliometrics within the same research assessment exercise.

cs.DL↗

arXiv2015

Regularized linear system identification using atomic, nuclear and kernel-based norms: the role of the stability constraint

Inspired by ideas taken from the machine learning literature, new regularization techniques have been recently introduced in linear system identification. In particular, all the adopted estimators solve a regularized least squares problem, differing in the nature of the penalty term assigned to the impulse response. Popular choices include atomic and nuclear norms (applied to Hankel matrices) as well as norms induced by the so called stable spline kernels. In this paper, a comparative study of estimators based on these different types of regularizers is reported. Our findings reveal that stable spline kernels outperform approaches based on atomic and nuclear norms since they suitably embed information on impulse response stability and smoothness. This point is illustrated using the Bayesian interpretation of regularization. We also design a new class of regularizers defined by "integral" versions of stable spline/TC kernels. Under quite realistic experimental conditions, the new estimators outperform classical prediction error methods also when the latter are equipped with an oracle for model order selection.

eess.SY↗

arXiv2011

Efficient Marginal Likelihood Computation for Gaussian Process Regression

In a Bayesian learning setting, the posterior distribution of a predictive model arises from a trade-off between its prior distribution and the conditional likelihood of observed data. Such distribution functions usually rely on additional hyperparameters which need to be tuned in order to achieve optimum predictive performance; this operation can be efficiently performed in an Empirical Bayes fashion by maximizing the posterior marginal likelihood of the observed data. Since the score function of this optimization problem is in general characterized by the presence of local optima, it is necessary to resort to global optimization strategies, which require a large number of function evaluations. Given that the evaluation is usually computationally intensive and badly scaled with respect to the dataset size, the maximum number of observations that can be treated simultaneously is quite limited. In this paper, we consider the case of hyperparameter tuning in Gaussian process regression. A straightforward implementation of the posterior log-likelihood for this model requires O(N^3) operations for every iteration of the optimization procedure, where N is the number of examples in the input dataset. We derive a novel set of identities that allow, after an initial overhead of O(N^3), the evaluation of the score function, as well as the Jacobian and Hessian matrices, in O(N) operations. We prove how the proposed identities, that follow from the eigendecomposition of the kernel matrix, yield a reduction of several orders of magnitude in the computation time for the hyperparameter optimization problem. Notably, the proposed solution provides computational advantages even with respect to state of the art approximations that rely on sparse kernel matrices.

stat.ML↗

arXiv2010

Client-server multi-task learning from distributed datasets

A client-server architecture to simultaneously solve multiple learning tasks from distributed datasets is described. In such architecture, each client is associated with an individual learning task and the associated dataset of examples. The goal of the architecture is to perform information fusion from multiple datasets while preserving privacy of individual data. The role of the server is to collect data in real-time from the clients and codify the information in a common database. The information coded in this database can be used by all the clients to solve their individual learning task, so that each client can exploit the informative content of all the datasets without actually having access to private data of others. The proposed algorithmic framework, based on regularization theory and kernel methods, uses a suitable class of mixed effect kernels. The new method is illustrated through a simulated music recommendation system.

cs.LG↗