3) Università degli Studi di Roma "La Sapienza"

Scientific Staff

Prof. Francesco Battaglia
Prof. Roberto Baragona
Prof. Enzo Orsingher
Prof. Stefano Fachin
Prof. Paolo Dell'Olmo

Francesco Battaglia graduated in statistics from the University of Rome and has been full professor of statistics since 1986 at the Universities of Cagliari and Rome "La Sapienza", and at Scuola Superiore della Pubblica Amministrazione, Rome. He has been head of the Department of Statistics, University La Sapienza, Rome, 1998- 2001.

Expertise of the Team

The team members belong to the Department of Statistics, Probability and Applied Statistics (Battaglia, Orsingher, Dell'Olmo), Department of National Accounting and Social Processes Analysis (Fachin), and Department of Sociology and Communication (Baragona). Baragona and Battaglia are statisticians, Orsingher is a probabilist, Fachin is an econometrician and Dell'Olmo is an operations researcher. They are all interested in heuristic optimization, and many of them have frequently participated in joint research. Focus will be on multi-disciplinary research, gathering expertise on computational statistics, time series analysis, econometric modelling and computational econometrics, and optimization algorithms. Some contributions to the topic of heuristic optimization have already been published. In particular, Baragona and Battaglia were among the first researchers to recognize the importance of heuristic optimization in time series analysis (Baragona and Battaglia, 2001). Later, their research focused on nonlinear time series analysis (Battaglia and Orfei, 2005), and the role of heuristic optimization in such a framework has been developed. A recent paper (Baragona, Battaglia and Cucina 2004) introduces a general class of nonlinear time series models (piecewise linear threshold autoregressive) and shows how the genetic algorithm may be used for identifying and estimating such models.

References

  • R. Baragona, F. Battaglia, and C. Calzini: “Genetic algorithms for the identification of additive and innovational outliers in time series”, Computational Statistics and Data Analysis, 2001.
  • R. Baragona, F. Battaglia, and D. Cucina: “Fitting piecewise linear threshold autoregressive models by means of genetic algorithms”, Computational Statistics and Data Analysis, 2004.
  • F. Battaglia and L. Orfei: “Outlier detection and estimation in non-linear time series”, Journal of Time Series Analysis, 2005.

Expertise in Training Young Scientists

Battaglia and Orsingher are members of the teachers’ team of the Ph.D. in Statistical Methodology at the Univ. La Sapienza Rome since its establishment (about 1985). This programme has 5-6 students each year. Dell'Olmo is head of the Ph.D. in Operations Research, Univ. La Sapienza Rome. Furthermore, Baragona, Battaglia and Dell'Olmo teach (the last two have also a position on the steering committee) in the second-level Master in Data Intelligence and Decision Strategies: a one-year post-graduate course started in 2003 with 25-30 students each
year.

Links within the Network

As part of the computational statistics international community, some team members have frequently collaborated with members of other teams. Baragona and Battaglia have contributed a paper to a special issue of Computational Statistics and Data Analysis on heuristic optimization edited by Winker and Gilli. Finally, Baragona and Battaglia will give an invited lecture at the Compstat 2006 conference that will be held in Rome, August 2006, in a specialised session on heuristic optimization organized by Winker.

Role of the Research Team

The research activity will concentrate on developing new heuristic optimization tools for building time series models for phenomena with strong nonlinearities and/or nonstationarities. Important examples are multi-regime models, conditional heteroskedasticity models, and the combination of both. Another relevant case is modelling abrupt modifications such as outliers or structural breaks. Application of heuristic optimization algorithms will be extended to the framework of multivariate non-Gaussian time series analysis. The research team will also contribute to the training and transfer of knowledge activities. Here also focus will be on multi-disciplinarity, offering a wide and integrated range of topics from statistics to econometrics, numerical optimization theory and operations research.