This Special Issue of ADAC aims to gather papers in different areas of research such as Statistics, Data Science, and Artificial Intelligence, to identify new models, methods, innovative procedures, and applications for producing high-quality data and timely statistics.Innovative multivariate statistical methods combined with data science and artificial intelligence techniques are now seen as useful approaches to modernize and speed up the data collection, analysis, and production of statistics useful to describe, forecast, and decide.This Special Issue was launched at the Conference of European Statistics Stakeholders, held in Rome at la Sapienza University.The conference was organized by ESAC, Eurostat, ECB, FENStatS, Istat, and the Bank of Italy and Department of Statistical Sciences, Sapienza University or Rome.The conference has seen the involvement of 571 participants with 273 attending the conference in Rome and 298 online.The conference successfully created conditions to connect researchers, users, and producers of statistics in order to systematically generate collaborations.This Special Issue 1 of volume 20 (2026) of the journal Advances in Data Analysis and Classification (ADAC) contains eleven articles that deal with Mixture of Regressions, Robust Gradient Boosting for Generalized Additive Models, Reconstructing the Foreign Direct Investments network, p-generalized probit and logistic regression, fuzzy clustering, kNN

We introduce a novel variable selection method for high-dimensional generalized linear models, called Randomized Smart Subset Selection (RS3). RS3 combines randomized block-wise model fitting with importance-weighted stochastic subset sampling to efficiently identify active predictors. Two extensions are proposed: (i) RS3+SCAD, which applies penalized regression with the SCAD penalty to a reduced set of candidate variables, improving computational efficiency while adaptively determining model size; and (ii) RS3+FPCS, a subsampling-based procedure for false positive control based on coefficient sign stability. Simulation studies demonstrate that RS3 and its variants achieve high true positive rates with low false discovery, even under moderate predictor correlation. Applications to gene expression data show that RS3+SCAD consistently identifies sparse, biologically interpretable gene sets, outperforming existing methods in terms of prediction accuracy and stability. MSC Codes: 62F40, 62J12, 62H30

This paper proposes a distance-based aggregation and consensus method for preference-approvals with potential applications in group decision-making, recommendation systems, and social choice theory. Preference-approvals are a type of preference data where individuals provide a list of approved alternatives in addition to a strict ranking. The proposed method aims to synthesise individual preference-approvals into a unified consensus representing the group’s collective view. The consensus is the preference-approval, which minimises the average distance within the whole set of voters. The performance of the proposed method is investigated through a sensitivity analysis on the weighting parameter, two simulation studies, and two real-data applications.

Abstract The integration of artificial intelligence (AI) into medical practices has emphasized the importance of ensuring the reliability of predictive confidence, as it influences decision-making and the efficacy of AI-driven solutions. This paper focuses on the utilization of a non-parametric machine learning method, with a particular focus on the computation of confidence scores for individual classifications. Unlike parametric approaches, non-parametric techniques like k-nearest neighbors (kNN) offer flexibility without imposing strict data distribution assumptions. Leveraging this flexibility, we propose a novel kNN approach that introduces confidence-awareness through a two-layered neighborhood analysis. The developed approach is intended to support the classical non-parametric kNN classifier by providing more reliable and trustworthy class probabilities. Experimental evaluations conducted on benchmark datasets as well as a de-identified clinical real-world Electronic Health Records (EHR) data table consisting of thousands of unique class labels demonstrate the effectiveness of our approach in enhancing both, prediction accuracy and certainty assessment.

Abstract Many point processes such as earthquakes or volcanic eruptions have incomplete records with the degree of incompleteness varying over time. For these point processes, the number of missing events between each pair of consecutively observed events can be a random variable that may depend on time, effecting the estimation of parameters or hazard. Such incomplete point processes can be modelled by compound renewal processes where the sum of renewal processes is a random variable because of random variable number of missing events. We propose shifted compound Poisson-Gamma and time-dependent shifted compound Poisson-Gamma renewal processes. Since the number of missing events can be regarded as an unobserved process, the proposed renewal processes are introduced to use in the framework of different types of homogeneous and inhomogeneous hidden Markov models to model the time-dependent variable number of missing events between each pair of consecutively observed events of incomplete point processes. Simulation experiments are employed to check the performance of proposed renewal processes with hidden Markov models. We apply the proposed models to the large magnitude explosive volcanic eruptions database to analyze the time-dependent incompleteness and demonstrate how we estimate the completeness of the record and the future hazard rate.