<s>
In	O
statistics	O
,	O
the	B-Application
bootstrap	I-Application
error-adjusted	O
single-sample	O
technique	O
(	O
BEST	O
or	O
the	O
BEAST	O
)	O
is	O
a	O
non-parametric	B-General_Concept
method	I-General_Concept
that	O
is	O
intended	O
to	O
allow	O
an	O
assessment	O
to	O
be	O
made	O
of	O
the	O
validity	O
of	O
a	O
single	O
sample	O
.	O
</s>
<s>
This	O
is	O
done	O
use	O
a	O
statistical	O
method	O
called	O
bootstrapping	B-Application
,	O
applied	O
to	O
previous	O
samples	O
that	O
are	O
known	O
to	O
be	O
valid	O
.	O
</s>
<s>
A	O
quantitative	O
approach	O
involves	O
BEST	O
along	O
with	O
a	O
nonparametric	B-General_Concept
cluster	B-Algorithm
analysis	I-Algorithm
algorithm	O
.	O
</s>
<s>
Multidimensional	O
standard	B-General_Concept
deviations	I-General_Concept
(	O
MDSs	O
)	O
between	O
clusters	O
and	O
spectral	O
data	O
points	O
are	O
calculated	O
,	O
where	O
BEST	O
considers	O
each	O
frequency	O
to	O
be	O
taken	O
from	O
a	O
separate	O
dimension	O
.	O
</s>
<s>
C	O
is	O
the	O
expectation	O
value	O
of	O
P	O
,	O
written	O
E(P )	O
,	O
and	O
B	O
is	O
a	O
bootstrapping	B-Application
distribution	O
called	O
the	O
Monte	B-Algorithm
Carlo	I-Algorithm
approximation	O
.	O
</s>
<s>
The	O
standard	B-General_Concept
deviation	I-General_Concept
can	O
be	O
found	O
using	O
this	O
technique	O
.	O
</s>
<s>
The	O
hyperline	O
from	O
C	O
to	O
X	O
gives	O
rise	O
to	O
the	O
skew	O
adjusted	O
standard	B-General_Concept
deviation	I-General_Concept
which	O
is	O
calculated	O
in	O
both	O
directions	O
of	O
the	O
hyperline	O
.	O
</s>
