[ Printable poster (pdf, 5.3M) ]
For abstract, location(s) and registration see
https://agenda.infn.it/event/20980/
 Lecture 1 (8 January)
 Slides:
References, links, etc.
 Lecture 2 (15 January)
 Slides:

wls_02.pdf
(with recommended home work^{(*)})
[^{(*)} Suggested addendum on the AIDS problem:
make a sensitivity study changing the
prior probability of infected by
±10%, ±20% and ±50%.
]
References, links, etc.
 C'è statistica e statistica, Scaffali di SxT, 30 marzo 2005 (pdf;
copia locale).
 When ISTAT tries to explain Bayesian reasoning:
 Teaching statistics in the physics curriculum.
Unifying and clarifying role of subjective probability,
AJP 67, issue 12 (1999) 12601268;
arXiv:physics/9908014
[Limited
to Sections IIV, for the moment]
 More lessons from the six box toy experiment,
arXiv:1701.01143.
 Constraints on the Higgs Boson Mass from Direct Searches and Precision Measurements, by GdA and G. Degrassi,
arXiv:hepph/9902226
updated: Constraining the Higgs boson mass through the combination of direct search and precision measurement results,
arXiv:hepph/0001269.
 Bayesian reasoning versus conventional statistics in High Energy Physics,
arXiv:physics/9811046
 Hugin
(Graphical User Interface; Samples)
 Readytouse models based on the sixboxes toy experiment:
 Try to edit the models
(within HUGIN), changing the probability
tables, adding nodes, etc..
 Try to write from scratch the (minimalist) model to solve
the AIDS problem, using the number suggested in the slides
for easy comparisons.
just two nodes
 Infected, with two possible states,
Yes and No;
 Analysis result, with two possible states,
Positive/ and Negative.
 Modify the previous model, using equiprobable
priors for Infected/NonInfected:
 compare the result with the those obtained
with (roughly) realistic priors;
 compare the result with the wrong one suggested
in the first lecture.
 Think then to the possible practical utility of
using equiprobable priors.
 An interesting classical example
is the so called `Asia':
(Indeed, a valid, for some aspects even better, alternative to Hugin is provided
by Netica, also
thank to the many available
tutorials
examples
whose interest goes beyond the specific package.)
 Lecture 3 (22 January)
 Slides:
 wls_03.pdf
(with recommended home work)
 Extra (very important!) problems in order
to understand meaning and role of probabilistic dependence/independence.
(We shall return on the topic in the context
of correlation coefficients)
 More on the distribution of the
product of the outcomes of two dice:
Alternative way to produce the histogram:
outcomes = as.vector(outer(1:6,1:6))
hist(outcomes, nc=40, freq=FALSE, col='cyan', xlab='x', ylab='f(x)')
 Then, here is out to make a random generator following the distribution:
sample(outcomes, 100, rep=TRUE)
(For help on the R functions you che use, e.g., ?sample, or search on the web.)
References, links, etc.
Now we are finally ready to analyze the
graphical model of the lectures poster.
Additional home work based on
the diagram:
 prove, just using physics arguments, that
λ = λ_{S} + λ_{B};
 then, using the reasoning used in the previous item,
draw the diagram in a different, more physical, way.
 Lecture 4 (29 January)

Slides:
References, links, etc.
 Lecture 5 (5 Febuary)
 Slides:
References, links, etc.
 Lecture 6 (12 February)
 Slides:
References, links, etc.