well guyz..
I have to implement an algorithm and it works as follows:

we need to take some different sized data sets ( numerical) , which has some objects and attributes
The algorithm first starts off with an empty set and it adds attributes thereafter
The dependency of each attribute is calculated. It is given by dividing the absolute value of those attributes which are certainly contained in the universal set over the absolute value of all the attributes ( or the universal set ) in the data set taken
Once the dependency is calculated for every attribute, the best candidate is chosen which has the highest dependency. Again the process is repeated and it continues until the dependency is equal to the consistency of the data set ( value = 1)

Well.. this is how the algorithm works..
Can someone tell me about how to go about starting or implementing this ?
(I'm new to the coding world.. so, if someone plz throw their ideas, ill be grateful)

raj chopra