The Best Ways to Learn from Data | School of Management and Decision sciences

Every day we generate more than 2 x 10^30 bytes of data coming from a variety of different sources, from sensors to web sites, from transaction records to digital pictures. This data are no longer manageable with known data management tools and data processing techniques. On the other side, enterprise performance management measurement as well as strategic decision making require the capability of implementing the best analysis in a very strict timeline. Thus, an unexpected number of business applications are available for enterprises and in the very next future the competition in several sectors will be determined by new capabilities in properly analyzing data.

The School of Management and Decisions Sciences has planned for this academic year a series of lectures held by distinguished speakers at international level. These lectures, taught by great worldwide recognized experts, will give a unique opportunity to executives and practitioners to be aware of the up-to-date management methodologies and tools. 

 

The Best Ways to Learn from Data

We introduce the “art” of learning from data offering an overview of the modern statistical techniques. The course will offer practical advises as well as methodological insights for social scientists and other applied statisticians. It will provide guidance into the process of building and evaluating models. Data analysis, variables transformations and graphical summaries of the results are emphasized.

We finally discuss some challenges that researchers have when trying to learn from real big data. We then present, through examples, various methods we use in our research.

 

The Speaker

 

Andrew Gelman is a professor of statistics and political science and director of the Applied Statistics Center at Columbia University. He has received the Outstanding Statistical Application award from the American Statistical Association, the award for best article published in the American Political Science Review, and the Council of Presidents of Statistical Societies award for outstanding contributions by a person under the age of 40. His books include Bayesian Data Analysis (with John Carlin, Hal Stern, and Don Rubin), Teaching Statistics: A Bag of Tricks (with Deb Nolan), Data Analysis Using Regression and Multilevel/Hierarchical Models (with Jennifer Hill), Red State, Blue State, Rich State, Poor State: Why Americans Vote the Way They Do (with David Park, Boris Shor, Joe Bafumi, and Jeronimo Cortina), and A Quantitative Tour of the Social Sciences (co-edited with Jeronimo Cortina).

Andrew has done research on a wide range of topics, including: why it is rational to vote; why campaign polls are so variable when elections are so predictable; the statistical challenges of estimating small effects; social network structure; arsenic in Bangladesh; radon in your basement; toxicology; medical imaging; and methods in surveys, experimental design, statistical inference, computation, and graphics. 

 

 

 

 

(TENTATIVE) SCHEDULE

Day one (Oct, 23)

14:30-15:00 Registration
15:00-15:10 Opening
15:10-17:00 Regression models for prediction
17:00-19:00 Statistical modeling and statistical practice

Day two (Oct, 24)

15:00-17:00 Statistical modeling and statistical practice
17:00-19:00 Two more complicated applications

Day three (Oct, 25)

15:00-17:00 Generalizing from sample to population
17:00-19:00 Checking and understanding fitted models
 

 

 

 

COURSE CONTENT

Click here to download a copy of the slides used during the course.

Regression models for prediction

  • Linear regression
  • Logistic regression
  • Choice models

Statistical modeling and statistical practice

  • Of beauty, sex, and power: statistical challenges in estimating small effects
  • Why we (usually) don't care about multiple comparisons
  • Some recent progress in simple statistical methods
  • Information visualization and statistical graphics

Two more complicated applications

  • ​Hierarchical modeling and prior information: An example from toxicology
  • Getting around detection limits in diluition assays

Generalizing from sample to population

  • Multilevel regression and poststratification
  • Survey weighting is a mess
  • Causal inference

Checking and understanding fitted models

  • Simulation-based model checking
  • Tools for understanding complex models
  • Checking and understanding fitted models

A recording of the module Regression models for prediction is available on YouTube at the following link

 

 

 

 

 

 

 

 

COURSE ATTENDEES

Martone Pietro Accenture
Tou Yun Xiang Pietro Accenture
Zobi Gianluca Accenture
Masserizzi  Marco Allianz
Abete  Tiziana Banca d'Italia
Cascarano  Michele Banca d'Italia
Crispino  Paolo Banca d'Italia
Di Salvatore  Antonietta Banca d'Italia
Matteucci Mirko Banca d'Italia
Paduanelli Stefania Banca d'Italia
Paparo Claudia Banca d'Italia
Scambelluri Valerio Banca d'Italia
Vercelli  Francesco Banca d'Italia
Gaudio Rossella Consorzio per il Distretto dell'Audiovisivo e dell'ICT
Castaldi Luigi FAO
Cerasa Fabiana FAO
Chin  Nancy FAO
Fabi Carola FAO
Hoffmeister Onno FAO
Kao Michael FAO
Moreno  Gladys FAO
Milizia  Francesco Regione Lazio
Guarini Maria Rosaria Sapienza - Università di Roma
Tognetti Gian Roberto Sapienza - Università di Roma
Brunetti Paolo Thrust Direct
Federico Bruno Università Di Cassino E Del Lazio Meridionale
Abate Giorgio Università Di Roma Tre
Pierini Andrea Università Di Roma Tre

Pierini Andrea

LOCATION

The course will be held in "Aula Gini". 
Facoltà di Ingegneria dell'Informazione, Informatica e Statistica
Sapienza – Università di Roma Piazzale Aldo Moro 5 - Roma 

(Click here for information on how to reach the room)

Course attendes will be provided with wireless Internet access through the Sapienza Wireless service. 

 

PRICES

€600 for all the course. A 50% discount is available for PhD students, master degree students and post-doc fellows. Please, contact the course secretariat if you wish to apply for a discount.

 

Information

To subscribe to the course please follow the instructions reported in the registration form. The submission deadline for registration requests is Monday, October 21 16:00. Please, contact us in case of late registration.

For information on programme, costs and discounts contact the course secretariat:

Sig.ra Cristina Puteo

formazione-dss@uniroma1.it