The search for genetic regions associated with complex diseases, such as cancer or Alzheimer's disease, is an important challenge that may lead to better diagnosis and treatment. The existence of millions of DNA variations, primarily single nucleotide polymorphisms (SNPs), may allow the fine dissection of such associations. However, studies seeking disease association are limited by the cost of genotyping SNPs. Therefore, it is essential to find a small subset of informative SNPs (tag SNPs) that may be used as good representatives of the rest of the SNPs. Several informative SNP selection methods have been developed. Our experiments compare favorably to all the prediction and statistical methods by selecting the least number of informative SNPs. We proposed algorithms for faster prediction which yielded acceptable trade off. We validated our results using the k-fold test and its many variations.