In many automated applications, large amount of data is collected every day and it is used to learn classifier as well as to make automated decisions. If that training data is biased towards or against certain entity, race, nationality, gender then mining model may leads to discrimination. This paper elaborate direct discrimination prevention method. The DRP algorithm modifies the original data set to prevent direct discrimination. Direct discrimination takes place when decisions are made based on discriminatory attributes those are specified by the user. The performance of this system is evaluated using measures MC, GC, DDPP, DPDM etc. Different discrimination measures can be used to discover the discrimination.