jhlee-visionkalman / accord

Automatically exported from code.google.com/p/accord
0 stars 0 forks source link

ID3 decision tree #64

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
when working with large data set the tree is degenerated and this exception is 
displayed:
'Index was outside the bounds of the array'.

What is the expected output? What do you see instead?

What version of the product are you using? On what operating system?
windows7 64bit accord.net 2.9.0

Please provide any additional information below.

Original issue reported on code.google.com by sabri...@gmail.com on 6 Sep 2013 at 2:21

GoogleCodeExporter commented 9 years ago
Thanks for reporting the issue! To aid in the debugging, could you please give 
a bit more details about the data set you are using, and how you are creating 
the tree? Sometimes this exception might occur when the number of discrete 
symbols in the dataset is different from the number of symbols specified when 
creating the tree.

Original comment by cesarso...@gmail.com on 6 Sep 2013 at 4:42

GoogleCodeExporter commented 9 years ago
Hi,
I worked on the "nursery database"
http://archive.ics.uci.edu/ml/machine-learning-databases/nursery/nursery.names

when launching the analysis with the C4.5 algorithm the tree is generated, but 
with the ID3 it returns an exception    

Original comment by sabri...@gmail.com on 9 Sep 2013 at 8:37

GoogleCodeExporter commented 9 years ago
[deleted comment]
GoogleCodeExporter commented 9 years ago
A few corrections and improvements have been made in the DecisionTree module. 
Should be available in the next release.

Original comment by cesarso...@gmail.com on 28 Nov 2013 at 8:57

GoogleCodeExporter commented 9 years ago
Those errors should have been fixed on 2.12; however, more testing might be 
needed before the issue can be closed.

Original comment by cesarso...@gmail.com on 6 Jan 2014 at 11:11

GoogleCodeExporter commented 9 years ago
[deleted comment]
GoogleCodeExporter commented 9 years ago
Hi,

I am using IRIS dataset from UCI repository for classification . As this is a 
large dataset, it throws the exception as mentioned above by Sabri...Is there 
any solution for this? Is the new modified version released? Or can I get the 
entire source code? Pls help

Original comment by rakshabn...@gmail.com on 16 May 2014 at 4:04

GoogleCodeExporter commented 9 years ago
I imported the iris dataset from UCI and it ran without problems. Please, can 
you confirm you are talking about the dataset on 
http://archive.ics.uci.edu/ml/datasets/Iris ?

Original comment by cesarso...@gmail.com on 27 May 2014 at 9:56

GoogleCodeExporter commented 9 years ago
[deleted comment]
GoogleCodeExporter commented 9 years ago
Thank you so much for your concern @cesaro ..I made 80% of the entire dataset 
(from the same link which you have sent )as the training data and the rest 20% 
test data just to check out the performance. But it threw the exception. when I 
took only 10 training points, it worked well. So the problem is with number of 
training data points?  Does the code handle large number of training points or 
does it cause any memory exception? I dont know. Pls help me asap.

Original comment by rakshabn...@gmail.com on 28 May 2014 at 3:47

GoogleCodeExporter commented 9 years ago
Ah sorry, I forgot one thing: the IRIS dataset contains continuous values, so 
it means it should be necessary to use the C4.5 algorithm instead. There had 
been some recent updates in the decision tree modules, which are only available 
in the development branch of the project. Can you confirm that you are using 
C4.5 and not ID3, and still get the exceptions? 

If that is the case, then this will likely be fixed on the next version of the 
framework. If you wish, you can also download the latest sources from the 
GitHub repository to test. They are available at

https://github.com/accord-net/framework/

Original comment by cesarso...@gmail.com on 28 May 2014 at 8:44