Data and Decision Analysis
(Academic Year 20182019)
This page contains material about the following courses:
 Data and Decision Analysis (code 2012400, 9 cfu)
 MSc in Engineering Management
 MSc in Computer and Automation  curriculum Robotics and
Automation
 Decision Analysis (code 2012411, 6 cfu)
 MSc in Computer and Automation  curriculum Information
Systems
 System Identification and Data Analysis (code 2000365, 9 cfu)
 MSc in Computer and Automation  curriculum Robotics and
Automation
 Data Analysis (code 2000666, 6 cfu)
 MSc in Computer and Automation  curriculum Information
Systems
 MSc in Applied Mathematics
 Analisi delle Decisioni (code 108147D, 6 cfu)
 Laurea Magistrale in Ingegneria Gestionale
Starting from AY 20172018:
 System Identification and Data Analysis (code 2000365) is
borrowed from Data and Decision Analysis (code 2012400)
 Decision Analysis (code 2012411), Data Analysis (code
2000666) and Analisi delle Decisioni (code 108147D) are borrowed
from Data and Decision Analysis (code 2012400)  part Decision
Analysis.
Index
News
 The results of the written test (25/9/2019) are available here.
 The results of the written test (6/9/2019) are available here.
Course organization
The course Data and Decision Analysis is split in two independent
components.
 Filtering Techniques (3 cfu, about 24 hours) is taught by A. Garulli. More
information about this component can be found at http://control.dii.unisi.it/ft.
 Decision Analysis (6 cfu, about 48 hours) is taught by A.
Giannitrapani. More information about this component can be found
below.
Instructor:
Lecture timetable:
Tuesday 
11.0013.00 
Room F

Thursday

10.0013.00 
Room F

Office hours:
Monday, 16.0017.00.
Tuesday, 14.0015.00.
Course outline:
Download the slides with
essential information about the course.
The Decision Analysis part (~48 hours) is
organized in two parts. Part I (~25 hours) provides an introduction to
decision problems and presents some methodologies for their analysis. Part
II (~23 hours) focuses on sequential decision problems which can be
modeled as Markov Decision Processes. At the end of the first part, there
is a midterm exam. At the end of the second part there is a project
assignment. See
How the exam works.
Course schedule:
March 5  April 4: 
lectures Part I

April 11: 
midterm exam (written test)



April 9  May 23: 
lectures Part II

June 6:

project assignment 
Contents
Part I
Required background
 Basics of probability calculus
Syllabus
 Introduction to decision problems
 Elements and structure of decision problems
 Influence diagrams (ID)
 Decision trees (DT)
 Review of probability calculus
 Solving ID and DT
 Risk profile of a strategy
 The value of information
 Risk attitudes and utility functions
Teaching material
 Textbook
 [MHD] "Making Hard Decisions"
 R. T. Clemen, T. Reilly
Pacific Grove: Duxbury, 2001
 Slides (courtesy of George Washington University)
 Survey of
applications
 J.L. Corner, C.W. Kirkwood, "Decision analysis applications in the
operations research literature, 19701989", Operations
Research, vol. 39, no. 2, pp. 206219, 1991. (fulltext
available from computers connected to the university network)
 D.L. Keefer, C.W. Kirkwood, J. L. Corner, "Perspective
on decision analysis applications, 1990–2001". Decision
analysis, vol. 1, no. 1, pp. 422, 2004.
 Example of decision analysis process
 Exercises
on probability calculus (in Italian)
 Past midterm exams (in Italian, English version available soon)
 Past project works (in Italian, English version available soon)
Part II
Required
background
 Basics of probability calculus and Markov chains
 Basics of Matlab programming
Syllabus
 Introduction to sequential decision problems
 Dynamic programming for deterministic models
 Markov decision processes
 Finite horizon problems
 Infinite horizon problems
 Risk averse control
Teaching material
 Slides  Matlab
code
 Matlab
tutorial by Gowtham Bellala at the EECS Departement of
University of Michigan
 Further readings
 [PDM]
"Processi decisionali markoviani" In: Modelli e metodi
decisionali in condizioni di incertezza e rischio. G. Ghiani, E.
Manni (2009). Cap. 5, pp. 171–201. McGrawHill. (in Italian)
 [MDP] “Markov
Decision Processes: Discrete Stochastic Dynamic Programming”
M. L. Puterman  WileyInterscience, 1994
 [DES] “Introduction to Discrete Event Dystems” (Ch. 7)
C. G. Cassandras, S. Lafortune, Springer, 2008
 D. J. White, "A
survey of applications of Markov decision processes", The
Journal of the Operational Research Society, vol. 44, no.
11, pp. 10731096. (fulltext available from computers connected to
the university network)
 Some applications of
Dynamic Programming to Computer Science problems
There are two ways to pass the exam:
 do the midterm exam and the project work
 do one of the six regular exams available during the academic
year and the project work
The midterm exam and the regular exams consists in a
written test containing three exercises on the topics covered in Part I.
During the test, the use of books, slides, notes, etc. is not allowed.
Schedule of written tests (up to July 2019):
 Midterm exam: April 11
 Regular exam: July 4
 Regular exam: July 24
The project work is a homework on the topics covered in Part II
and it requires the use of Matlab. The project work can be carried out
individually or in groups of 34 students. Students have to turn in the
homework solution within one week after the project assignment.
Schedule of project assignments (up to July 2019)
 June 6 (deadline June 12)
 June 27 (deadline July 3)
 July 11 (deadline July 17)
Students must sign up to the lists available at https://segreteriaonline.unisi.it/
Until July 2019, any combination of the previous written test dates and
project assignment dates is allowed. Starting from September 2019,
students must contact the instructor before the written test date and ask
for the assignment of the project.
Lectures
 Tuesday, March 5
 Introduction to decision problems [MHD  Ch. 1]
 Thursday, March 7
 Elements of a decision problem [MHD
 Ch. 2]
 Tuesday, March 12
 Influence diagrams [MHD  Ch. 3]
 Thursday, March 14
 Decision trees [MHD  Ch. 3]
 Review of probability calculus [MHD
 Ch. 7]
 Tuesday, March 19
 Maximum expected monetary value. Solving decision trees [MHD
 Ch. 4]
 Thursday, March 21
 Value of information [MHD 
Ch. 12]
 Tuesday, March 26
 Risk profile of a strategy [MHD  Ch. 4]
 Risk attitude and utility functions [MHD 
Ch. 13]
 Thursday, March 28
 Maximum expected utility. Certainty equivalent and risk premium.
Utility function assessment [MHD 
Ch. 13]
 Tuesday, April 2
 Thursday, April 4
 Tuesday, April 9
 Thursday, April 11
 Monday, April 15
 Tuesday, April 30
 Thursday, May 2
 Monday, May 6
 Tuesday, May 7
 Example: Home Energy Management System (HEMS) [ Matlab
code ]
 Thursday, May 16
 Tuesday, May 21
 Thursday, May 23
 Tuesday, May 28
 Tuesday, June 4
 Thursday, June 6
Please report errors or problems to: giannitrapani at
dii.unisi.it