Maybe theres is a confusion for the review between samples and features 

How many samples/rows did you use?

Thanks Dr. Ahmad,

The number of samples is varied between 100 to 5000. But many papers in the literature stated that the main challenge when dealing with classification problem is when using a dataset with few samples while huge number of features (like the case in medical datasets). In one dataset we have 73 samples and 325 features.

Please have a look over this website. All datasets are the same (small number of samples and large number of features).

http://featureselection.asu.edu/datasets.php

Thanks a lot

Yyes their is some datasets with high dimensions.

Please see the attached paper : Feature subset selection using differential volutionand and a wheel based search strategy.

I think you should use some of these datasets, at least two of them.

Good Luck.

Sure, both are challenging. In my view, if your target is to test the generalisation of your classifier the more samples you have the better judgement you get. If the target is to test your dimension reduction method, then the more features you have the better judgement you get. But of course at the end both are important. 

Thank you for the datasets

You are welcome

Thanks a lot Wassila Guendouzi for your reply and the paper.

I have some of these datasets, but the number of features is less than 1000. I'll try to find one or two which contain more than 1000 features.

I'd appreciate if you could send me some ready to use datasets with more than 1000 features.

Regards,

Majdi

y are welocome, you can download these dataset from this link : https://archive.ics.uci.edu/ml/datasets.html

it is the UCI repository, for example : Dorothea, Amazon CommerceReviews, Dexter or Dorothea datasets.

Regards.