We discuss two-sample problems, specifically, how to implement unified statistical methods to help application of statistical methods which can be used for both discrete and continuous data so that we can give insights to dataset. The unified statistical tools are based on concepts of mid-distribution, design of score functions, component correlations, comparison density and exponential model. Our approach to two-sample problem is to use density estimation(comparison density estimation). Specifying the density function gives a full and natural description of the data. This approach is functional in the sense that the parameters to be estimated are probability density functions. Compared with other statistical tools for two-sample problems such as t-test or Wilcoxon rank-sum test, density estimation makes us use the data more fully, which is essential in data analysis. Also our approach using comparison density gives a unified tool for several statistical tools. We can solve ! problems of comparison of two samples, multi samples and goodness of fit test through comparison density concept.