of couse, you should use R with the data.table package (for large datasets)

Dear Mr Gajendra, in my humble opinion... there is no best software (advantages and disadvantages). Sometimes the combination of independent functions from various softwares provides a better result. I suggest to you to take a look on the following softwares:

• Weka (http://www.cs.waikato.ac.nz/ml/weka/)
• R (http://cran.r-project.org/)
• Orange (http://orange.biolab.si/)
• Knime (https://www.knime.org/)
• RapidMiner Community Edition (https://rapidminer.com)

Best regards.

R, Python or Java. I prefer R for the data analysis. It is open source software.

I think the choice solely depend upon the type of data. For example, if you have data from next generation sequencing machines then for instance python may help. if data is of metabolomics or networks, graph databases such as Neo4j. R is definitely good option for medium size dataset but can only used after preprocessing data coming from any throughput technology. If you want to model data and use machine learning approaches then WEKA is the most preferred choice. You may also have a look at HDF5 (http://www.hdfgroup.org/HDF5/).

Thanks for suggestions, our group is already using R, Weka, , SVMlight, SNNS, RAPID miner for developing prediction methods. One of the major problem is speed, if I build model on large number of patterns than techniques like SVM take huge time. This BIG data is new terminology, we heard recently. I am interested whether their are software tools that can mine large data in reasonable time. I means is their any tool specifically designed to develop model on huge data. For example Hadoop is specially developed  for managing BIG data. 

i have experience since two years with big dataset for analysis and i would like to recomend R for it!

recommended program open source for big data is LIBSVM Accelerated with GPU using the CUDA Framework

R is the best and user friendly software...

H2O is another good option:

http://0xdata.com/h2o/

I would suggest R as well, since it has a range of different packages and a good community behind it. 

I guess with the help Cloudera ...CDH3 software list Hadoop would be more useful.

Here are a number of research papers at NIH on Hadoop in bio-informatics

http://www.ncbi.nlm.nih.gov/pubmed?term=hadoop

Here is another good source:

http://abhishek-tiwari.com/post/mapreduce-and-hadoop-algorithms-in-bioinformatics-papers

Don't rule out Python, mllib in spark, madlib, mahout, SAS etc.  All depends on the problem you are trying to solve...

Currently, the best statistical analysis and mining package is R and the best infrastructure that can handle big data is Hadoop. So combining both together can make sense of big data! I suggest use "RHadoop" package..

You may like to read about how to install it in this link:

https://bighadoop.wordpress.com/2013/02/25/r-and-hadoop-data-analysis-rhadoop/

"RHadoop" uses the steaming feature recently embedded in Hadoop (MR2) and it turns the R code implicitly into efficient MapReduce code that run easily on HDFS.

https://bighadoop.wordpress.com/2013/02/25/r-and-hadoop-data-analysis-rhadoop/

Dear Gajendra,

Your can consider 'stream mining' software like MOA: http://moa.cms.waikato.ac.nz/

Kind regards,

Olivier PARISOT

If you are looking for batch-level processing of the data, then Hadoop stack is your best choice. However, you are looking for more real-time data mining, then Spark or datastax (http://www.datastax.com/) is a better route.

Thanks everyone for your excellent suggestions.

try this if your data are heavy tailed: http://en.wikipedia.org/wiki/Head/tail_Breaks

Use Python.