My problem is to run an Experiment of Classification using Weka (Weka Experimenter) for a time period of 1 hour. Total number of samples to be supplied for the duration of One hour is 50000 samples. Starting with 5000 samples, I need to increase samples by 5000 after every 6 minute i.e. Initially supplied dataset should be of 5000 samples, after 6 Minutes, it should be of 10000. After 12 Minutes, it should be of 15000 and so on.
I am not sure about incremental dataset, as i experienced weka it need to give full dataset once and apply any classification algorithm.
Please state why you not giving full dataset once?
Hi,
In fact, WEKA allows applying the standard classification as well as the incremental classification. The available WEKA's methods perform normal incremental classification (instance-by-instance) with an incremental evaluator (if used) that could accomplish learning even less than one hour. Hence, for your scenario, you need to write the code yourself.
HTH.
Samer
Hi,
I would say one solution is to implement the logic -as suggested. Also, the problem sounds like it includes data streams. I would recommend having a look at MOA (Massive Online Analysis) which is like Weka and can be used for data stream mining.
https://moa.cms.waikato.ac.nz/
Hi Ercan,
Even MOA doesn't apply the proposed scenario, the only possibility is to write a code.
Cheers,
Samer
Dear Prof. Singh,
You have to have a mathematical model developed scientifically prior to experiment and then accordingly set up the codes for the genetic algorithmic first step to fit the data samples by some sampling theory and set down a criterion by which to decide on convergence criterion and in next iteration subdivide the variables and the codes till you get epsilon convergence and so on. You can take alook at www.econometricsociety.org/[email protected] website for some of our reswults on String Classified and ihtegrated genetic algorithms results (Mallick, Hmaburger, Mallick (2016,2017)).
Soumitra K. Mallick
for Soumitra K Mallick, Nick Hamburger, Sandipan Mallick
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running