Version 7.0, released 14 May 2016
We are pleased to announce the release of Version 7.0
of AgenaRisk. As usual Professional licence holders, with a current
annual subscription, are entitled to a
free upgrade (simply follow the download instructions provided in the
email that was sent with your license file). Version 7.0 includes
the following important new functionality:
learning from data and expert judgment (see Section 10 of the Manual).
This enables you to automatically learn node probability tables from a
spreadsheet file and incorporate as much or as little expert judgment
as you wish. In all cases learning can be performed even when
there are ‘missing data’ in the spreadsheet. The learning
process is performed using an Expectation-Maximisation (EM) algorithm
and AgenaRisk currently supports table learning for both Boolean and
Labelled nodes. See this simple tutorial
video of this new functionality.
sum analysis (see Section 11 of the Manual).
This enables you to automatically compute the compund 'total
value' distribution where there are events with a frequency
distribution, and where each of these events has an associated severity
distribution. The severity distribution might represent an individual
loss or profit event, the time to failure of a component, or some other
variable. The frequency would be the number of times these events might
be expected to occur over a defined period of time. For example, an
insurer may hold a portfolio of insurance policies where any of these
may involve a claim – the frequency of claims will be uncertain
and the severity of each claim will be uncertain too. What the insurer
would like to know is the total claim amount as a function of the
frequency of claims and the severity of each claim. Mathematically this
problem is solved using a process called convolution, which AgenaRisk
does automatically by performing arithmetical operations on simulation
example models including "Paternity testing", "House
buying", "DNA testing with errors", "How to transform a continuous node
to a ranked node", "Beta distribution parameter learning", "Table
Learning", "Compound Sum".
There are also improvements to efficiency and many minor bug fixes.
There are some changes to the Free and Lite versions (in terms of
restrictions in functionality). See the new full
list of version comparison and features).
Remember to join the new Users Group
if you have not already done so. As well as answering user questions,
this is where we now post tips and tricks, video tutorials, and