Dear All,
I have been struggling for some time to implement the Maximum Entropy
(or Generalized Maximum Entropy) method in a range of software
platforms.
The intention is to use these to reconstruct transaction values in an
input-output matrix subject to the constraints of known row and column
totals and in a way which maximises the Shannon entropy function. This
method follows Golan, A.; Judge, G.; et al.(1996).Maximum Entropy
Econometrics: robust estimation with limited data. Chichester, John
Wiley& Sons (amongst one of many).
The intention is then to use these reconstructed values as a conjugate
Dirichlet prior for Multinomial data from another source in a larger
model.
The question is - is there any way I can do the Maximum Entropy, Gen Max
Ent, or some comparable procedure for data reconstruction within
WinBUGS?
Alternatively - has anybody else managed to implement Max Ent (for
something other than image reconstruction) in another package
(preferably Mathematica)?
Any help greatly appreciated.
David.
Dr David Shipworth
Coordinator: External Environments Research Division
School of Construction Management and Engineering
The University of Reading
Ph: (direct) +44 (0)118 378 7177
Ph: (office) +44 (0)118 378 8201
Fax:(office) +44 (0)118 931 3856
Further information on Maximum Entropy reconstruction:
The aim is to reconstruct values pjk in the matrix below so that:
1. The columns sum to the given column totals
2. The rows sum to the given row totals
3. The sum of the pjk values across the rows equals 1
4. The sum of the pjk values down the columns equals 1
5. The values selected for the pjk maximise the 'entropy function' which
takes the general form: -
In the general case:
n*m table
Col 1
Col k
Col m
Row Marginals
Row 1
pjk
pjk
pjk
TotRow 1
Row j
pjk
pjk
pjk
TotRow j
Row n
pjk
pjk
pjk
TotRow n
Column Marginals
TotCol 1
TotCol k
TotCol m
S S
Summary from below:
MaxEnt equation =
Constraint equation =
Summation equation =
Constraint: 0 * p(jk) * 1, p * R
-------------------------------------------------------------------
This list is for discussion of modelling issues and the BUGS software.
For help with crashes and error messages, first mail [log in to unmask]
To mail the BUGS list, mail to [log in to unmask]
Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html
Please do not mail attachments to the list.
To leave the BUGS list, send LEAVE BUGS to [log in to unmask]
If this fails, mail [log in to unmask], NOT the whole list