Want to protect your cyber security and still get fast solutions? Ask a secure question today.Go Premium

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1576
  • Last Modified:

Looking for a Decision Tree algorithm in C# or .NET?

I am looking for a good decision tree algorithm written in C# or at least .NET, so far my searches have not turned up much fruit. Could anyone suggest one?
0
Poolcorp
Asked:
Poolcorp
  • 6
  • 6
1 Solution
 
TommySzalapskiCommented:
I'm not sure what you are looking for. A decision tree is just a bunch of nested if statements. You don't need any fancy algorithms.
0
 
PoolcorpAuthor Commented:
Well I would like a documented object model, or something perhaps that can build a d-tree off of a dataset.
0
 
TommySzalapskiCommented:
How is the dataset structured? It sould be straight forward but will be very dependent on how the data is set up. You might just need a loop and some ifs, you might need a stack, or you could need a binary tree with a parser. What specifically are you trying to do?
0
Concerto Cloud for Software Providers & ISVs

Can Concerto Cloud Services help you focus on evolving your application offerings, while delivering the best cloud experience to your customers? From DevOps to revenue models and customer support, the answer is yes!

Learn how Concerto can help you.

 
PoolcorpAuthor Commented:
Basically I am using it as a forecasting tool, where based on a set of data with 20 columns and say 1 million rows a decision tree is built where it points to a specific column as the output. Then taking another set of data applying it to the decision tree it picks the expected output.
0
 
TommySzalapskiCommented:
I think you want a neural network, not a decision tree.
Decision trees are for things that have distinct yes and no paths. If you are trying to guess at one value (column number) based on other values, then you want a neural network or some other type of prediction scheme.
0
 
PoolcorpAuthor Commented:
Actually i am basing my paper off the differences between a neural network vs a decision tree and their respective accuracy regarding the type of data I am using. I have my NN built already but i was hoping to save some time on the decision tree side of things.
0
 
TommySzalapskiCommented:
Okay. I get it. So you are trying to build a classifier using discrete decisions and compare that to the NN. Are you wanting to find the optimal decision tree?
Are all your decisions binary (yes/no;  0/1; <5/>=5) or are they buckets (1 to 5, 6 to 10, 10 to 15)?

If they are all binary, then you could just test all possible ordering of the decisions and pick the best (should take some time). If they are not binary, then you'll need to figure out the optimal bucketing as well unless you are using static buckets.
0
 
PoolcorpAuthor Commented:
They will not be binary no, and the optimal decision tree is not a issue really. I was thinking about going from a largest to smallest approach, meaning take the values have the least amount of change as the base of the tree and go from there

0
 
TommySzalapskiCommented:
That is a very unique approach at a classifier which is why you were unable to find algorithms for it on the web.
So you are basically going to look at classifying each data point with only one feature. Then pick the feature that does the best job and set that as the first feature. Then repeat the process with the remaining features to select the next one until all have been selected. Right? That seems like a reasonable approach and should be fairly simple to implement.
0
 
PoolcorpAuthor Commented:
That is the general gist i thought of
0
 
TommySzalapskiCommented:
If you do a web search for "Classification Trees" or "Regression Trees" you will get some more complex ideas. I believe those are the terms you were originally looking for.
0
 
PoolcorpAuthor Commented:
Ok, Thanks!
0

Featured Post

Keep up with what's happening at Experts Exchange!

Sign up to receive Decoded, a new monthly digest with product updates, feature release info, continuing education opportunities, and more.

  • 6
  • 6
Tackle projects and never again get stuck behind a technical roadblock.
Join Now