Hi,

Uncertain data mining is a hot topic. In frequent pattern mining, there have been several people working on uncertain itemset mining. I have made a small search online and found some source code and datasets for uncertain itemset mining, that I will share with you:

• Java source code of the MBP and IMBP algorithms + synthetic dataset (T10I4D100K) + real datasets (retail_dat) : http://code.google.com/p/imbp-programe-dataset/
• Java source code of the UApriori algorithm in the SPMF data mining framework: http://www.philippe-fournier-viger.com/spmf/
• C++ source code of the p-fp, p-apriori, TODIS and p-AR algorithms + synthetic dataset (T25I10D500) and real dataset (Accident). http://i.cs.hku.hk/~lwsun/codes/kdd10/

I will add more later. If you have any other link to uncertain itemset mining source code or datasets, please post them below.

Philippe



Edited 3 time(s). Last edit at 10/15/2012 03:30PM by webmasterphilfv.

Hi philfv, that is grate. Pls let me have a taste of it.

So it can be be new way for all the data mining sub field? like we can have uncertain dataset for the sequential mining, graph mining, etc.....

Yes. Since a few year, I have seen many work on uncertain data.

But one thing that I have noticed is that there is not a lot of real datasets available for uncertain itemset mining. Therefore many researchers uses synthetic datasets instead of real data. I don,t know if this has changed recently though...

Hi philfv,
I am working on uncertain itemset mining problems. while implemnting these algorithms,how can we create uncertain database in coding level(java). The programe which is provided in "UApriori algorithm" in the SPMF data mining framework allows the already predefined probability value. but, how can we associate probability value for each items in the real dataset? do you have any code snippets in java or any idea regarding this?

Thanks,
Hema

Hi,

You would need to read the papers about uncertain algorithms. It is a long time that i have read them so i have forogt. But I think that they use some statistic distributions to generate the probability. To implement this, it should not be very complicated. You just need to understand the statistic distribution and to use a kind of random generator.

To do that in SPMF, you could modify the method "loadfile" in the source code for UApriori. When loading the file, you could write some code to dynamically assigns probabilities according to the distributions.

Another way would be to write a simple program that read a file, add some probabilities and then write the file as output. Then you could give the file to SPMF with the generated probabilities.

This is just some ideas.

But if you find a way to do it, you could share the code on the forum. That would be great ;-)

Philippe

hii,

i want u-apiriori for uncertain data code....plz help me...

Hi,

You can get the Java source code of UApriori algorithm as part of the SPMF data mining library. Just go to download page and follows the instructions.

Philippe



Edited 1 time(s). Last edit at 09/11/2013 05:11AM by webmasterphilfv.