<s>
In	O
decision	B-Algorithm
tree	I-Algorithm
learning	I-Algorithm
,	O
Information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
is	O
a	O
ratio	O
of	O
information	B-Algorithm
gain	I-Algorithm
to	O
the	O
intrinsic	O
information	O
.	O
</s>
<s>
Information	B-Algorithm
Gain	I-Algorithm
is	O
also	O
known	O
as	O
Mutual	O
Information	O
.	O
</s>
<s>
Information	B-Algorithm
gain	I-Algorithm
is	O
the	O
reduction	O
in	O
entropy	O
produced	O
from	O
partitioning	O
a	O
set	O
with	O
attributes	O
and	O
finding	O
the	O
optimal	O
candidate	O
that	O
produces	O
the	O
highest	O
value	O
:	O
</s>
<s>
The	O
information	B-Algorithm
gain	I-Algorithm
is	O
equal	O
to	O
the	O
total	O
entropy	O
for	O
an	O
attribute	O
if	O
for	O
each	O
of	O
the	O
attribute	O
values	O
a	O
unique	O
classification	O
can	O
be	O
made	O
for	O
the	O
result	O
attribute	O
.	O
</s>
<s>
This	O
in	O
turn	O
is	O
the	O
intrinsic	O
value	O
that	O
the	O
random	O
variable	O
possesses	O
and	O
will	O
be	O
used	O
to	O
remove	O
the	O
bias	O
in	O
the	O
Information	B-General_Concept
Gain	I-General_Concept
Ratio	I-General_Concept
calculation	O
.	O
</s>
<s>
The	O
information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
is	O
the	O
ratio	O
between	O
the	O
information	B-Algorithm
gain	I-Algorithm
and	O
the	O
Split	O
Information	O
value	O
:	O
</s>
<s>
Using	O
the	O
table	O
above	O
,	O
one	O
can	O
find	O
the	O
Entropy	O
,	O
Information	B-Algorithm
Gain	I-Algorithm
,	O
Split	O
Information	O
,	O
and	O
Information	B-General_Concept
Gain	I-General_Concept
Ratio	I-General_Concept
for	O
each	O
variable	O
(	O
Outlook	O
,	O
Temperature	O
,	O
Humidity	O
,	O
and	O
Wind	O
)	O
.	O
</s>
<s>
Using	O
the	O
above	O
tables	O
,	O
one	O
can	O
deduce	O
that	O
Outlook	O
has	O
the	O
highest	O
information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
.	O
</s>
<s>
Humidity	O
was	O
found	O
to	O
have	O
the	O
highest	O
information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
.	O
</s>
<s>
Since	O
the	O
Play	O
values	O
are	O
either	O
all	O
"	O
NO	O
"	O
or	O
"	O
YES	O
"	O
,	O
the	O
information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
value	O
will	O
be	O
equal	O
to	O
1	O
.	O
</s>
<s>
Also	O
,	O
now	O
that	O
one	O
has	O
reached	O
the	O
end	O
of	O
the	O
variable	O
chain	O
with	O
Wind	O
being	O
the	O
last	O
variable	O
left	O
,	O
they	O
can	O
build	O
an	O
entire	O
root	O
to	O
leaf	O
node	O
branch	O
line	O
of	O
a	O
decision	B-Algorithm
tree	I-Algorithm
.	O
</s>
<s>
Once	O
finished	O
with	O
reaching	O
this	O
leaf	O
node	O
,	O
one	O
would	O
follow	O
the	O
same	O
procedure	O
for	O
the	O
rest	O
of	O
the	O
elements	O
that	O
have	O
yet	O
to	O
be	O
split	O
in	O
the	O
decision	B-Algorithm
tree	I-Algorithm
.	O
</s>
<s>
This	O
set	O
of	O
data	O
was	O
relatively	O
small	O
,	O
however	O
,	O
if	O
a	O
larger	O
set	O
was	O
used	O
,	O
the	O
advantages	O
of	O
using	O
the	O
information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
as	O
the	O
splitting	O
factor	O
of	O
a	O
decision	B-Algorithm
tree	I-Algorithm
can	O
be	O
seen	O
more	O
.	O
</s>
<s>
Information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
biases	O
the	O
decision	B-Algorithm
tree	I-Algorithm
against	O
considering	O
attributes	O
with	O
a	O
large	O
number	O
of	O
distinct	O
values	O
.	O
</s>
<s>
Example	O
:	O
Suppose	O
that	O
we	O
are	O
building	O
a	O
decision	B-Algorithm
tree	I-Algorithm
for	O
some	O
data	O
describing	O
a	O
business	O
's	O
customers	O
.	O
</s>
<s>
Information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
is	O
used	O
to	O
decide	O
which	O
of	O
the	O
attributes	O
are	O
the	O
most	O
relevant	O
.	O
</s>
<s>
This	O
attribute	O
has	O
a	O
high	O
information	B-Algorithm
gain	I-Algorithm
,	O
because	O
it	O
uniquely	O
identifies	O
each	O
customer	O
.	O
</s>
<s>
Although	O
information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
solves	O
the	O
key	O
problem	O
of	O
information	B-Algorithm
gain	I-Algorithm
,	O
it	O
creates	O
another	O
problem	O
.	O
</s>
<s>
Information	B-Algorithm
gain	I-Algorithm
's	O
shortcoming	O
is	O
created	O
by	O
not	O
providing	O
a	O
numerical	O
difference	O
between	O
attributes	O
with	O
high	O
distinct	O
values	O
from	O
those	O
that	O
have	O
less	O
.	O
</s>
<s>
Example	O
:	O
Suppose	O
that	O
we	O
are	O
building	O
a	O
decision	B-Algorithm
tree	I-Algorithm
for	O
some	O
data	O
describing	O
a	O
business	O
's	O
customers	O
.	O
</s>
<s>
Information	B-Algorithm
gain	I-Algorithm
is	O
often	O
used	O
to	O
decide	O
which	O
of	O
the	O
attributes	O
are	O
the	O
most	O
relevant	O
,	O
so	O
they	O
can	O
be	O
tested	O
near	O
the	O
root	O
of	O
the	O
tree	O
.	O
</s>
<s>
This	O
attribute	O
has	O
a	O
high	O
information	B-Algorithm
gain	I-Algorithm
,	O
because	O
it	O
uniquely	O
identifies	O
each	O
customer	O
,	O
but	O
we	O
do	O
not	O
want	O
to	O
include	O
it	O
in	O
the	O
decision	B-Algorithm
tree	I-Algorithm
:	O
deciding	O
how	O
to	O
treat	O
a	O
customer	O
based	O
on	O
their	O
credit	O
card	O
number	O
is	O
unlikely	O
to	O
generalize	O
to	O
customers	O
we	O
have	O
n't	O
seen	O
before	O
.	O
</s>
<s>
Information	B-General_Concept
gain	I-General_Concept
ratio	I-General_Concept
's	O
strength	O
is	O
that	O
it	O
has	O
a	O
bias	O
towards	O
the	O
attributes	O
with	O
the	O
lower	O
number	O
of	O
distinct	O
values	O
.	O
</s>
<s>
Below	O
is	O
a	O
table	O
describing	O
the	O
differences	O
of	O
Information	B-Algorithm
Gain	I-Algorithm
and	O
Information	B-General_Concept
Gain	I-General_Concept
Ratio	I-General_Concept
when	O
put	O
in	O
certain	O
scenarios	O
.	O
</s>
