<s>
mlpy	B-Application
is	O
a	O
Python	B-Language
,	O
open-source	B-Application
,	O
machine	O
learning	O
library	O
built	O
on	O
top	O
of	O
NumPy/SciPy	O
,	O
the	O
GNU	B-Application
Scientific	I-Application
Library	I-Application
and	O
it	O
makes	O
an	O
extensive	O
use	O
of	O
the	O
Cython	B-Application
language	O
.	O
</s>
<s>
mlpy	B-Application
provides	O
a	O
wide	O
range	O
of	O
state-of-the-art	O
machine	O
learning	O
methods	O
for	O
supervised	O
and	O
unsupervised	O
problems	O
and	O
it	O
is	O
aimed	O
at	O
finding	O
a	O
reasonable	O
compromise	O
among	O
modularity	O
,	O
maintainability	O
,	O
reproducibility	O
,	O
usability	O
and	O
efficiency	O
.	O
</s>
<s>
mlpy	B-Application
is	O
multiplatform	O
,	O
it	O
works	O
with	O
Python	B-Language
2	O
and	O
3	O
and	O
it	O
is	O
distributed	O
under	O
GPL3	O
.	O
</s>
<s>
Suited	O
for	O
general-purpose	O
machine	O
learning	O
tasks	O
,	O
mlpy	B-Application
's	O
motivating	O
application	O
field	O
is	O
bioinformatics	O
,	O
i.e.	O
</s>
<s>
Many	O
classification	O
and	O
regression	O
algorithms	O
are	O
endowed	O
with	O
an	O
internal	O
feature	O
ranking	O
procedure	O
:	O
in	O
alternative	O
,	O
mlpy	B-Application
implements	O
the	O
I-Relief	O
algorithm	O
.	O
</s>
<s>
Methods	O
for	O
feature	O
list	O
analysis	O
(	O
for	O
example	O
the	O
Canberra	O
stability	O
indicator	O
)	O
,	O
data	O
resampling	O
and	O
error	O
evaluation	O
are	O
provided	O
,	O
together	O
with	O
different	O
clustering	O
analysis	O
methods	O
(	O
Hierarchical	O
,	O
Memory-saving	O
Hierarchical	O
,	O
k-means	B-Algorithm
)	O
.	O
</s>
