Hi,
This is in response to Peter and Adreas' queries about the differences between
CART and C4.5/C5.0
1. Splits in CART are restricted to binary splits (a C4.5/C5.0 D-Tree is m-ary)
2. Differences between C4.5/C5.0 and CART include differences in:
a. splitting criteria
b. the pruning method
c. and the way missing values are handled.
A detailed description of the above are available in [1]
3. C4.5 (and hence C5.0) is in fact and improved version of the ID3 and takes
care of some of the issues that ID3 does not take care of [2]
a. Avoiding overfitting the data
Determining how deeply to grow a decision tree.
b. Reduced error pruning.
c. Rule post-pruning.
d. Handling continuous attributes.
e.g., temperature
e. Choosing an appropriate attribute selection measure.
f. Handling training data with missing attribute values.
g. Handling attributes with differing costs.
h. Improving computational efficiency.
Additional Reading:
1. http://ms1psz.etf.rs/vezbe/Data%20Mininig/05%20DECISION%20TREES.pdf
2. http://www.stat.wisc.edu/~loh/class.pdf
( 2. above summarizes the differences quite well)
To be absolutely honest, I'm currently an undergrad. I read about the C4.5
algorithm a few months back and am myself making quite an effort to understand
some of the terms above. (i.e I don't fully understand some of the terms in the
reading material and the references cited). I hope to although, by the middle of
the application period have understood most of them.
Thanks,
Vikram Kamath
References.
[1] ai.stanford.edu/~ronnyk/treesHB.pdf
[2] http://www2.cs.uregina.ca/~dbd/cs831/notes/ml/dtrees/c4.5/tutorial.html
------------------------------------------------------------------------------
Try before you buy = See our experts in action!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-dev2
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
