For many classification tasks, I apply the scikit-learn library in python...

http://scikit-learn.org/

It offers you several classifiers, including ensemble techniques (gradient boosted tree classifiers etc.). As it is sometimes necessary, to implement your own code, I would still recommend to use such a tested library as proven-ground to start with.

Dear Priya,

to implement the many classification task, i usually use my own code due to every data type has different process. in my opinion, ensemble classifiier is better because there are comparisons among them so that it makes proper decision. Comparison can be done by multiplying to their own weight/gain and then get the best value.

nice to discuss with you :)

Dear Priya Gs,

In this case, you can apply Committee Machine, a supervised learning based on divide-and-conquer principles. It means learning tasks will be distributed among various experts( e.g. SVM, ANN, k-NN etc) and final output will be combined based on a strategy that maximizes the objective.

So, some well-know combiner strategies are as follows,

1. Ensemble averaging

2. Boosting.

3. Mixture of ex[erets

I have attached a flow-diagram to implement your idea. In addition, You can review this book?

HAYKIN, S. Neural Networks - A Comprehensive Foundation. Second edition. Pearson Prentice Hall: 1999.

Thanks

Regards,

M M Manjurul Islam

https://www.researchgate.net/profile/M_M_Manjurul_Islam/contributions

Dear Pryia,

You can use different hybrid methods, and it is related to your dataset and problem. I suggest you to study this concept firstly. Afterward, you can select a right way to implement it. Even though you can use the weka or other similar applications, some powerful methods must be implemented by you like Mixture Of Experts. Please read the below links which are related to ensemble learning techniques.

http://ieeexplore.ieee.org/abstract/document/1688199/?reload=true

https://books.google.com/books?hl=en&lr=&id=B4TZtp7X82oC&oi=fnd&pg=PR5&dq=Kuncheva+combining+pattern+classifiers&ots=IgX9orS2Jp&sig=oNLxDsC5jr_dXBnE6QEK2-vblwk#v=onepage&q=Kuncheva%20combining%20pattern%20classifiers&f=false