<s>
C4.5	B-Algorithm
is	O
an	O
algorithm	O
used	O
to	O
generate	O
a	O
decision	B-Algorithm
tree	I-Algorithm
developed	O
by	O
Ross	O
Quinlan	O
.	O
</s>
<s>
C4.5	B-Algorithm
is	O
an	O
extension	O
of	O
Quinlan	O
's	O
earlier	O
ID3	B-Algorithm
algorithm	I-Algorithm
.	O
</s>
<s>
The	O
decision	B-Algorithm
trees	I-Algorithm
generated	O
by	O
C4.5	B-Algorithm
can	O
be	O
used	O
for	O
classification	O
,	O
and	O
for	O
this	O
reason	O
,	O
C4.5	B-Algorithm
is	O
often	O
referred	O
to	O
as	O
a	O
statistical	B-General_Concept
classifier	I-General_Concept
.	O
</s>
<s>
In	O
2011	O
,	O
authors	O
of	O
the	O
Weka	B-Language
machine	O
learning	O
software	O
described	O
the	O
C4.5	B-Algorithm
algorithm	I-Algorithm
as	O
"	O
a	O
landmark	O
decision	B-Algorithm
tree	I-Algorithm
program	O
that	O
is	O
probably	O
the	O
machine	O
learning	O
workhorse	O
most	O
widely	O
used	O
in	O
practice	O
to	O
date	O
"	O
.	O
</s>
<s>
It	O
became	O
quite	O
popular	O
after	O
ranking	O
#1	O
in	O
the	O
Top	O
10	O
Algorithms	O
in	O
Data	B-Application
Mining	I-Application
pre-eminent	O
paper	O
published	O
by	O
Springer	O
LNCS	O
in	O
2008	O
.	O
</s>
<s>
C4.5	B-Algorithm
builds	O
decision	B-Algorithm
trees	I-Algorithm
from	O
a	O
set	O
of	O
training	O
data	O
in	O
the	O
same	O
way	O
as	O
ID3	B-Algorithm
,	O
using	O
the	O
concept	O
of	O
information	O
entropy	O
.	O
</s>
<s>
Each	O
sample	O
consists	O
of	O
a	O
p-dimensional	O
vector	O
,	O
where	O
the	O
represent	O
attribute	O
values	O
or	O
features	B-Algorithm
of	O
the	O
sample	O
,	O
as	O
well	O
as	O
the	O
class	O
in	O
which	O
falls	O
.	O
</s>
<s>
At	O
each	O
node	O
of	O
the	O
tree	O
,	O
C4.5	B-Algorithm
chooses	O
the	O
attribute	O
of	O
the	O
data	O
that	O
most	O
effectively	O
splits	O
its	O
set	O
of	O
samples	O
into	O
subsets	O
enriched	O
in	O
one	O
class	O
or	O
the	O
other	O
.	O
</s>
<s>
The	O
splitting	O
criterion	O
is	O
the	O
normalized	O
information	B-Algorithm
gain	I-Algorithm
(	O
difference	O
in	O
entropy	O
)	O
.	O
</s>
<s>
The	O
attribute	O
with	O
the	O
highest	O
normalized	O
information	B-Algorithm
gain	I-Algorithm
is	O
chosen	O
to	O
make	O
the	O
decision	O
.	O
</s>
<s>
The	O
C4.5	B-Algorithm
algorithm	I-Algorithm
then	O
recurses	O
on	O
the	O
partitioned	O
sublists	O
.	O
</s>
<s>
When	O
this	O
happens	O
,	O
it	O
simply	O
creates	O
a	O
leaf	O
node	O
for	O
the	O
decision	B-Algorithm
tree	I-Algorithm
saying	O
to	O
choose	O
that	O
class	O
.	O
</s>
<s>
None	O
of	O
the	O
features	B-Algorithm
provide	O
any	O
information	B-Algorithm
gain	I-Algorithm
.	O
</s>
<s>
In	O
this	O
case	O
,	O
C4.5	B-Algorithm
creates	O
a	O
decision	O
node	O
higher	O
up	O
the	O
tree	O
using	O
the	O
expected	O
value	O
of	O
the	O
class	O
.	O
</s>
<s>
Again	O
,	O
C4.5	B-Algorithm
creates	O
a	O
decision	O
node	O
higher	O
up	O
the	O
tree	O
using	O
the	O
expected	O
value	O
.	O
</s>
<s>
In	O
pseudocode	B-Language
,	O
the	O
general	O
algorithm	O
for	O
building	O
decision	B-Algorithm
trees	I-Algorithm
is	O
:	O
</s>
<s>
For	O
each	O
attribute	O
a	O
,	O
find	O
the	O
normalized	O
information	B-Algorithm
gain	I-Algorithm
ratio	O
from	O
splitting	O
on	O
a	O
.	O
</s>
<s>
Let	O
a_best	O
be	O
the	O
attribute	O
with	O
the	O
highest	O
normalized	O
information	B-Algorithm
gain	I-Algorithm
.	O
</s>
<s>
J48	O
is	O
an	O
open	B-License
source	I-License
Java	B-Language
implementation	O
of	O
the	O
C4.5	B-Algorithm
algorithm	I-Algorithm
in	O
the	O
Weka	B-Language
data	B-Application
mining	I-Application
tool	O
.	O
</s>
<s>
C4.5	B-Algorithm
made	O
a	O
number	O
of	O
improvements	O
to	O
ID3	B-Algorithm
.	O
</s>
<s>
Handling	O
both	O
continuous	O
and	O
discrete	O
attributes	O
-	O
In	O
order	O
to	O
handle	O
continuous	O
attributes	O
,	O
C4.5	B-Algorithm
creates	O
a	O
threshold	O
and	O
then	O
splits	O
the	O
list	O
into	O
those	O
whose	O
attribute	O
value	O
is	O
above	O
the	O
threshold	O
and	O
those	O
that	O
are	O
less	O
than	O
or	O
equal	O
to	O
it	O
.	O
</s>
<s>
Handling	O
training	O
data	O
with	O
missing	O
attribute	O
values	O
-	O
C4.5	B-Algorithm
allows	O
attribute	O
values	O
to	O
be	O
marked	O
as	O
?	O
</s>
<s>
Pruning	O
trees	O
after	O
creation	O
-	O
C4.5	B-Algorithm
goes	O
back	O
through	O
the	O
tree	O
once	O
it	O
's	O
been	O
created	O
and	O
attempts	O
to	O
remove	O
branches	O
that	O
do	O
not	O
help	O
by	O
replacing	O
them	O
with	O
leaf	O
nodes	O
.	O
</s>
<s>
C5.0	O
offers	O
a	O
number	O
of	O
improvements	O
on	O
C4.5	B-Algorithm
.	O
</s>
<s>
Smaller	O
decision	B-Algorithm
trees	I-Algorithm
-	O
C5.0	O
gets	O
similar	O
results	O
to	O
C4.5	B-Algorithm
with	O
considerably	O
smaller	O
decision	B-Algorithm
trees	I-Algorithm
.	O
</s>
<s>
Support	O
for	O
boosting	B-Algorithm
-	O
Boosting	B-Algorithm
improves	O
the	O
trees	O
and	O
gives	O
them	O
more	O
accuracy	O
.	O
</s>
<s>
Winnowing	O
-	O
a	O
C5.0	O
option	O
automatically	O
winnows	B-Algorithm
the	O
attributes	O
to	O
remove	O
those	O
that	O
may	O
be	O
unhelpful	O
.	O
</s>
<s>
Source	O
for	O
a	O
single-threaded	O
Linux	O
version	O
of	O
C5.0	O
is	O
available	O
under	O
the	O
GNU	B-License
General	I-License
Public	I-License
License	I-License
(	O
GPL	B-License
)	O
.	O
</s>
