Pyfmtools
Pyfmtools provides various tools for handling fuzzy measures, calculating various indices, Choquet and Sugeno integrals, as well as fitting fuzzy measures to empirical data. This package is designed for Python , but it also includes the C++ source files and a user manual. Chapter 2 of the user manual provides some background on fuzzy measures. A more detailed overview can be found in [4, 5, 12, 16] and references therein. Chapter 3 of the user manual outlines computational methods used to fit fuzzy measures to empirical data. The description of the programming library pyfmtools is given in Chapter 4. Examples of its usage are provided in Section 4.6. To cite pyfmtools package, use references [2–6,21–24].
New in version 4
Random generation of fuzzy measures of different types, including k-additive, k-interactive, supermodular and submodular, sparse representation of k- additive fuzzy measures.
We added in version 3
We added the concept of K-interactive fuzzy measures, and 4 methods of fitting K-interactive fuzzy measures from data based on linear program- ming. K-interactive fuzzy measures significantly reduce the computational complexity. We also added fitting fuzzy measures in marginal contribution representation and using maximal chains method, which fits only the values directly identifiable from the data. This method is useful for small data sets.
Fitting fuzzy measures in marginal contribution representation allows simple sub and supermodularity constraints, which can now be enforced.
See functions fittingKinteractive, fittingKinteractiveAuto, fittingKinter- activeMC, fittingKinteractiveMarginal, fittingKinteractiveMarginalMC.
We added calculation of new non-additivity and bipartition interaction indices. See functions Bipartition, BipartitionBanzhaf, NonadditivityIndex, NonadditivityIndexMob.
We added in version 2
We added fitting K-maxitive and K-tolerant fuzzy measures, based on linear and mixed integer programming. See functions fittingktolerant and fittingK- maxitive.
We added a method for fitting sub-modular fuzzy measures reported in [3]. Supermodular fuzzy measure can also be fit by using duality: construct dual data set, fit a sub-modular fuzzy measure and then compute its dual. See function FuzzyMeasureFitLP.
We added an extra requirement of preservation of output ordering. See function FuzzyMeasureFitLP.
Fixed many warnings in the lpsolve code.
Documentation
Installation
To install type:
$ pip install pyfmtools
Usage of the library
How to use the library with Python only
import pyfmtools as pyfm
Follow these steps in your Python code to use the library:
- Initialize resources
- Use library function(s)
- Free resources
Example how to initialize resources
n=3<br>
env = pyfm.fm_init( n)<br>
Example how to free resources
pyfm.fm_free( env)<br>
Usage of library functions
To implement a function follow these steps:
- Initialize input arrays.
- Initialize input parameters.
- Call wrapper function.
- Evaluate output parameters.
Example
import pyfmtools as pyfm
n=3
env = pyfm.fm_init( n)
k = 2
Mob =[0.0,0.3,0.5,-0.2,0.4,0.1,-0.2,0.1]
pnm = pyfm.NonmodularityIndexMobkadditive(Mob, k, env)
print( "k: ", k)
print( "Mob: ", Mob)
print( "nonmodularity indices: ", pnm)
pyfm.fm_free( env)
Parameters
Input parameters:
See input parameter list in user manual
Output parameters:
See output parameter list in user manual
Test
To unit test type:
$ test/test_wrapper.py
How to use the library utilizing NumPy and CFFI
from _pyfmtools import flib, lib
Follow these steps in your Python code to use the library:
- Initialize resources
- Use library function(s)
- Free resources
Example how to initialize resources
n=3
env=ffi.new( "struct fm_env *")
fm.py_fm_init( n, env)
Example how to free resources
fm.py_fm_free( env)<br>
Usage of library functions
To implement a function follow these steps:
- Initialise input parameters
- Initialise input arrays.
- Covert input arrays to CFFI.
- Initialise output arrays.
- Covert output arrays to CFFI.
- Call C function.
Parameters
Input parameters:
See input parameter list in user manual
Output parameters:
See output parameter list in user manual
Example
import numpy as np
from _pyfmtools import ffi,lib as fm
n=4
fm.py_fm_init(n, env);
ti=1
v = np.zeros(env.m,float);
pv = ffi.cast("double *", v.ctypes.data);
vb = np.zeros(env.m,float);
pvb = ffi.cast("double *", vb.ctypes.data);
size = fm.py_generate_fm_2additive_concave(ti,n,pv)
print( "2-additive concave FM in Mobius and its length (n=4)")
print( v)
print( "has ", size, " nonzero parameters ")
fm.py_fm_free( env);
Test
To unit test type:
$ test/test_no_wrapper.py
Release files for pyfmtools 5.4.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| pyfmtools-5.4.0.tar.gz | 537.0 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| pyfmtools-5.4.0-cp314-cp314-macosx_14_0_arm64.whl | CPython 3.14 | CPython 3.14 | macOS 14.0+ ARM64 | Details |
Total release size: 1.0 MB
Release files / pyfmtools-5.4.0.tar.gz
| Download URL | pyfmtools-5.4.0.tar.gz |
|---|---|
| Size | 537.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
b7e7288e51caa95d6e28c50e7e0bb43d4f1720ce762121e0d6a88573a87495ca
|
|
BLAKE2b-256 checksum How to use checksums |
cabe476900a9d7aea71824434065fd078ea5619fef093342fc658dd5680f768a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.7
|
Release files / pyfmtools-5.4.0-cp314-cp314-macosx_14_0_arm64.whl
| Download URL | pyfmtools-5.4.0-cp314-cp314-macosx_14_0_arm64.whl |
|---|---|
| Size | 507.8 kB |
| Tags | CPython 3.14 macOS 14.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
56f459a03b91269fe360065d1ea9e74179fdda14a36d2bd23db93b1b2201c087
|
|
BLAKE2b-256 checksum How to use checksums |
c56129e0b0740fb9c84197b6700b9d5fd4e38335215e661b1d78cfec0c78f8de
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.7
|
