<s>
Automatic	B-Algorithm
clustering	I-Algorithm
algorithms	I-Algorithm
are	O
algorithms	O
that	O
can	O
perform	O
clustering	O
without	O
prior	O
knowledge	O
of	O
data	O
sets	O
.	O
</s>
<s>
In	O
contrast	O
with	O
other	O
cluster	B-Algorithm
analysis	I-Algorithm
techniques	O
,	O
automatic	B-Algorithm
clustering	I-Algorithm
algorithms	I-Algorithm
can	O
determine	O
the	O
optimal	O
number	O
of	O
clusters	O
even	O
in	O
the	O
presence	O
of	O
noise	O
and	O
outlier	O
points	O
.	O
</s>
<s>
Automated	O
selection	O
of	O
k	O
in	O
a	O
K-means	B-Algorithm
clustering	I-Algorithm
algorithm	I-Algorithm
,	O
one	O
of	O
the	O
most	O
used	O
centroid-based	O
clustering	B-Algorithm
algorithms	I-Algorithm
,	O
is	O
still	O
a	O
major	O
problem	O
in	O
machine	O
learning	O
.	O
</s>
<s>
The	O
most	O
accepted	O
solution	O
to	O
this	O
problem	O
is	O
the	O
elbow	B-General_Concept
method	I-General_Concept
.	O
</s>
<s>
It	O
consists	O
of	O
running	O
k-means	B-Algorithm
clustering	I-Algorithm
to	O
the	O
data	O
set	O
with	O
a	O
range	O
of	O
values	O
,	O
calculating	O
the	O
sum	O
of	O
squared	O
errors	O
for	O
each	O
,	O
and	O
plotting	O
them	O
in	O
a	O
line	O
chart	O
.	O
</s>
<s>
Another	O
method	O
that	O
modifies	O
the	O
k-means	B-Algorithm
algorithm	I-Algorithm
for	O
automatically	O
choosing	O
the	O
optimal	O
number	O
of	O
clusters	O
is	O
the	O
G-means	O
algorithm	O
.	O
</s>
<s>
Thus	O
,	O
k	O
is	O
increased	O
until	O
each	O
k-means	B-Algorithm
center	O
's	O
data	O
is	O
Gaussian	O
.	O
</s>
<s>
Connectivity-based	O
clustering	O
or	O
hierarchical	B-Algorithm
clustering	I-Algorithm
is	O
based	O
on	O
the	O
idea	O
that	O
objects	O
have	O
more	O
similarities	O
to	O
other	O
nearby	O
objects	O
than	O
to	O
those	O
further	O
away	O
.	O
</s>
<s>
Although	O
hierarchical	B-Algorithm
clustering	I-Algorithm
has	O
the	O
advantage	O
of	O
allowing	O
any	O
valid	O
metric	O
to	O
be	O
used	O
as	O
the	O
defined	O
distance	O
,	O
it	O
is	O
sensitive	O
to	O
noise	O
and	O
fluctuations	O
in	O
the	O
data	O
set	O
and	O
is	O
more	O
difficult	O
to	O
automate	O
.	O
</s>
<s>
Methods	O
have	O
been	O
developed	O
to	O
improve	O
and	O
automate	O
existing	O
hierarchical	B-Algorithm
clustering	I-Algorithm
algorithms	O
such	O
as	O
an	O
automated	O
version	O
of	O
single	O
linkage	O
hierarchical	B-Algorithm
cluster	I-Algorithm
analysis	I-Algorithm
(	O
HCA	O
)	O
.	O
</s>
<s>
Information	O
gathered	O
from	O
HCA	O
,	O
automated	O
and	O
reliable	O
,	O
can	O
be	O
resumed	O
in	O
a	O
dendrogram	B-Application
with	O
the	O
number	O
of	O
natural	O
clusters	O
and	O
the	O
corresponding	O
separation	O
,	O
an	O
option	O
not	O
found	O
in	O
classical	O
HCA	O
.	O
</s>
<s>
BIRCH	B-Algorithm
(	O
balanced	O
iterative	O
reducing	O
and	O
clustering	O
using	O
hierarchies	O
)	O
is	O
an	O
algorithm	O
used	O
to	O
perform	O
connectivity-based	O
clustering	O
for	O
large	O
data-sets	O
.	O
</s>
<s>
It	O
is	O
regarded	O
as	O
one	O
of	O
the	O
fastest	O
clustering	B-Algorithm
algorithms	I-Algorithm
,	O
but	O
it	O
is	O
limited	O
because	O
it	O
requires	O
the	O
number	O
of	O
clusters	O
as	O
an	O
input	O
.	O
</s>
<s>
Therefore	O
,	O
new	O
algorithms	O
based	O
on	O
BIRCH	B-Algorithm
have	O
been	O
developed	O
in	O
which	O
there	O
is	O
no	O
need	O
to	O
provide	O
the	O
cluster	O
count	O
from	O
the	O
beginning	O
,	O
but	O
that	O
preserves	O
the	O
quality	O
and	O
speed	O
of	O
the	O
clusters	O
.	O
</s>
<s>
The	O
main	O
modification	O
is	O
to	O
remove	O
the	O
final	O
step	O
of	O
BIRCH	B-Algorithm
,	O
where	O
the	O
user	O
had	O
to	O
input	O
the	O
cluster	O
count	O
,	O
and	O
to	O
improve	O
the	O
rest	O
of	O
the	O
algorithm	O
,	O
referred	O
to	O
as	O
tree-BIRCH	O
,	O
by	O
optimizing	O
a	O
threshold	O
parameter	O
from	O
the	O
data	O
.	O
</s>
<s>
For	O
this	O
,	O
other	O
algorithms	O
have	O
been	O
developed	O
,	O
like	O
MDB-BIRCH	O
,	O
which	O
reduces	O
super	O
cluster	O
splitting	O
with	O
relatively	O
high	O
speed	O
.	O
</s>
<s>
Unlike	O
partitioning	O
and	O
hierarchical	O
methods	O
,	O
density-based	O
clustering	B-Algorithm
algorithms	I-Algorithm
are	O
able	O
to	O
find	O
clusters	O
of	O
any	O
arbitrary	O
shape	O
,	O
not	O
only	O
spheres	O
.	O
</s>
<s>
The	O
density-based	O
clustering	B-Algorithm
algorithm	I-Algorithm
uses	O
autonomous	O
machine	O
learning	O
that	O
identifies	O
patterns	O
regarding	O
geographical	O
location	O
and	O
distance	O
to	O
a	O
particular	O
number	O
of	O
neighbors	O
.	O
</s>
<s>
The	O
fastest	O
method	O
is	O
DBSCAN	B-Algorithm
,	O
which	O
uses	O
a	O
defined	O
distance	O
to	O
differentiate	O
between	O
dense	O
groups	O
of	O
information	O
and	O
sparser	O
noise	O
.	O
</s>
<s>
Lastly	O
,	O
the	O
method	O
OPTICS	B-Algorithm
creates	O
a	O
reachability	O
plot	O
based	O
on	O
the	O
distance	O
from	O
neighboring	O
features	O
to	O
separate	O
noise	O
from	O
clusters	O
of	O
varying	O
density	O
.	O
</s>
<s>
The	O
Automatic	O
Local	O
Density	O
Clustering	B-Algorithm
Algorithm	I-Algorithm
(	O
ALDC	O
)	O
is	O
an	O
example	O
of	O
the	O
new	O
research	O
focused	O
on	O
developing	O
automatic	O
density-based	O
clustering	O
.	O
</s>
<s>
Clustering	B-Algorithm
algorithms	I-Algorithm
artificially	O
generated	O
are	O
compared	O
to	O
DBSCAN	B-Algorithm
,	O
a	O
manual	O
algorithm	O
,	O
in	O
experimental	O
results	O
.	O
</s>
