Wolfram Language Paclet Repository
Community-contributed installable additions to the Wolfram Language
AntonAntonov`DimensionReducers`
NonNegativeMatrixFactorization  |
|
| | ||||
Details and Options
Examples
(6)
Basic Examples
(1)
Create a random integer matrix:
In[1]:=
SeedRandom[7]mat=RandomInteger[10,{4,3}];MatrixForm[mat]
Out[1]=
RandomGeneratorState
|
Out[1]//MatrixForm=
4 | 7 | 4 |
10 | 8 | 8 |
5 | 3 | 4 |
5 | 4 | 5 |
Compute the NNMF factors:
In[2]:=
{W,H}=
[mat,2];Row[{MatrixForm[W],MatrixForm[H]}]
 | NonNegativeMatrixFactorization |
Out[2]=
0.688539 | 0.0354316 |
0.625313 | 0.766099 |
0.196628 | 0.466821 |
0.310216 | 0.440358 |
5.42607 | 10.0437 | 5.45225 |
8.36432 | 2.19475 | 6.34806 |
Here is the matrix product of the obtained factors:
In[3]:=
MatrixForm[W.H]
Out[3]//MatrixForm=
4.03242 | 6.99325 | 3.97901 |
9.80089 | 7.96186 | 8.2726 |
4.97155 | 2.99943 | 4.03547 |
5.36655 | 4.0822 | 4.48679 |
Note that elementwise relative errors between the original matrix and reconstructed matrix are small:
In[4]:=
MatrixForm[Round[(W.H-mat)/mat,0.001]]
Out[4]//MatrixForm=
0.008 | -0.001 | -0.005 |
-0.02 | -0.005 | 0.034 |
-0.006 | 0. | 0.009 |
0.073 | 0.021 | -0.103 |
Scope
(1)
Options
(2)
Applications
(1)
Properties & Relations
(1)
 |
|
""