<s>
The	O
Iris	B-Language
flower	I-Language
data	I-Language
set	I-Language
or	O
Fisher	B-Language
's	I-Language
Iris	I-Language
data	I-Language
set	O
is	O
a	O
multivariate	B-General_Concept
data	I-General_Concept
set	O
used	O
and	O
made	O
famous	O
by	O
the	O
British	O
statistician	O
and	O
biologist	O
Ronald	O
Fisher	O
in	O
his	O
1936	O
paper	O
The	O
use	O
of	O
multiple	O
measurements	O
in	O
taxonomic	O
problems	O
as	O
an	O
example	O
of	O
linear	B-General_Concept
discriminant	I-General_Concept
analysis	I-General_Concept
.	O
</s>
<s>
It	O
is	O
sometimes	O
called	O
Anderson	B-Language
's	I-Language
Iris	I-Language
data	I-Language
set	I-Language
because	O
Edgar	O
Anderson	O
collected	O
the	O
data	O
to	O
quantify	O
the	O
morphologic	O
variation	O
of	O
Iris	O
flowers	O
of	O
three	O
related	O
species	O
.	O
</s>
<s>
The	O
data	B-General_Concept
set	I-General_Concept
consists	O
of	O
50	O
samples	O
from	O
each	O
of	O
three	O
species	O
of	O
Iris	O
(	O
Iris	O
setosa	O
,	O
Iris	O
virginica	O
and	O
Iris	O
versicolor	O
)	O
.	O
</s>
<s>
Four	O
features	B-Algorithm
were	O
measured	O
from	O
each	O
sample	O
:	O
the	O
length	O
and	O
the	O
width	O
of	O
the	O
sepals	O
and	O
petals	O
,	O
in	O
centimeters	O
.	O
</s>
<s>
Based	O
on	O
the	O
combination	O
of	O
these	O
four	O
features	B-Algorithm
,	O
Fisher	O
developed	O
a	O
linear	O
discriminant	O
model	O
to	O
distinguish	O
the	O
species	O
from	O
each	O
other	O
.	O
</s>
<s>
Originally	O
used	O
as	O
an	O
example	O
data	B-General_Concept
set	I-General_Concept
on	O
which	O
Fisher	B-General_Concept
's	I-General_Concept
linear	I-General_Concept
discriminant	I-General_Concept
analysis	O
was	O
applied	O
,	O
it	O
became	O
a	O
typical	O
test	O
case	O
for	O
many	O
statistical	B-General_Concept
classification	I-General_Concept
techniques	O
in	O
machine	O
learning	O
such	O
as	O
support	B-Algorithm
vector	I-Algorithm
machines	I-Algorithm
.	O
</s>
<s>
The	O
use	O
of	O
this	O
data	B-General_Concept
set	I-General_Concept
in	O
cluster	B-Algorithm
analysis	I-Algorithm
however	O
is	O
not	O
common	O
,	O
since	O
the	O
data	B-General_Concept
set	I-General_Concept
only	O
contains	O
two	O
clusters	O
with	O
rather	O
obvious	O
separation	O
.	O
</s>
<s>
This	O
makes	O
the	O
data	B-General_Concept
set	I-General_Concept
a	O
good	O
example	O
to	O
explain	O
the	O
difference	O
between	O
supervised	O
and	O
unsupervised	O
techniques	O
in	O
data	B-Application
mining	I-Application
:	O
Fisher	B-General_Concept
's	I-General_Concept
linear	I-General_Concept
discriminant	I-General_Concept
model	O
can	O
only	O
be	O
obtained	O
when	O
the	O
object	O
species	O
are	O
known	O
:	O
class	O
labels	O
and	O
clusters	O
are	O
not	O
necessarily	O
the	O
same	O
.	O
</s>
<s>
Nevertheless	O
,	O
all	O
three	O
species	O
of	O
Iris	O
are	O
separable	O
in	O
the	O
projection	O
on	O
the	O
nonlinear	O
and	O
branching	O
principal	B-Application
component	I-Application
.	O
</s>
<s>
The	O
data	B-General_Concept
set	I-General_Concept
is	O
approximated	O
by	O
the	O
closest	O
tree	O
with	O
some	O
penalty	O
for	O
the	O
excessive	O
number	O
of	O
nodes	O
,	O
bending	O
and	O
stretching	O
.	O
</s>
<s>
For	O
each	O
node	O
the	O
pie	B-Application
diagram	I-Application
of	O
the	O
projected	O
points	O
is	O
prepared	O
.	O
</s>
<s>
Therefore	O
,	O
the	O
three	O
species	O
of	O
Iris	O
(	O
Iris	O
setosa	O
,	O
Iris	O
virginica	O
and	O
Iris	O
versicolor	O
)	O
are	O
separable	O
by	O
the	O
unsupervising	O
procedures	O
of	O
nonlinear	O
principal	B-Application
component	I-Application
analysis	I-Application
.	O
</s>
<s>
The	O
dataset	B-General_Concept
contains	O
a	O
set	O
of	O
150	O
records	O
under	O
five	O
attributes	O
-	O
sepal	O
length	O
,	O
sepal	O
width	O
,	O
petal	O
length	O
,	O
petal	O
width	O
and	O
species	O
.	O
</s>
<s>
The	O
iris	B-Language
data	I-Language
set	I-Language
is	O
widely	O
used	O
as	O
a	O
beginner	O
's	O
dataset	B-General_Concept
for	O
machine	O
learning	O
purposes	O
.	O
</s>
<s>
The	O
dataset	B-General_Concept
is	O
included	O
in	O
R	B-Language
base	O
and	O
Python	B-Language
in	O
the	O
machine	O
learning	O
library	O
scikit-learn	B-Application
,	O
so	O
that	O
users	O
can	O
access	O
it	O
without	O
having	O
to	O
find	O
a	O
source	O
for	O
it	O
.	O
</s>
<s>
Several	O
versions	O
of	O
the	O
dataset	B-General_Concept
have	O
been	O
published	O
.	O
</s>
<s>
The	O
example	O
R	B-Language
code	I-Language
shown	O
below	O
reproduce	O
the	O
scatterplot	B-Application
displayed	O
at	O
the	O
top	O
of	O
this	O
article	O
:	O
</s>
