Efficient Computational and Statistical Methods for, but not limited to, the Social Science

Decanato - Facoltà di scienze informatiche

Data: 7 Novembre 2017 / 09:30 - 10:30

USI Lugano Campus, room A-13, Red building (Via G. Buffi 13)

Speaker: Stefano M. Iacus
  University of Milan, Italy
Date: Tuesday, November 7, 2017
Place: USI Lugano Campus, room A-13, Red building (Via G. Buffi 13)
Time: 09:30-10:30

 

Abstract:

In this talk we discuss three topics in which efficient computational and statistical methods play a relevant role. At first, we discuss a problem of data matching for large scale databases where the aim is the efficient treatment effect estimation in non-experimental studies, a typical issue of economic and political science studies, which is becoming more and more common in medicine and health studies. 

Then we move to the problem of text analysis in social media data and discuss an efficient algorithm which is not based on NLP or machine learning approach. Applications to monitoring of terrorism recruiting and political elections will be presented. 

Finally, we discuss an object oriented computational environment which allows for efficient simulation and inference of stochastic differential equations driven by Gaussian or Fractional Gaussian noise or Jump processes of Lévy type. This environment also allows for counting processes (Poisson, Hawkes, etc) and continuous counterparts of traditional econometric models, like COGARCH and CARMA. Applications to finance and social media analysis will be discussed.

 

Biography:

Stefano M. Iacus, PhD, is full professor of Statistics the Department of Economics, Management and Quantitative Methods at the University of Milan. He has been a member of the R Core Team (1999-2014) for the development of the R statistical environment and now member of the R Foundation for Statistical Computing. He is also a co-founder of VOICES from the Blogs, a startup company of the University of Milan, specialised in Social Media and Big Data Analysis.

His research interests include inference for stochastic processes, simulation, quantitative finance, computational statistics, causal inference and sentiment analysis. 

 

Host: Prof. Antonio Carzaniga