An Affinity Propagation Clustering Algorithm for Mixed Numeric and Categorical Datasets

Algorithm 2

Pseudocode of the proposed algorithm.

Calculate the similarity matrix and preferences as input of AP algorithm.

for each numeric attribute in dataset   do

 Calculate the significance of attribute using the method in Section 3.1;

end for

for each categorical attribute in dataset   do

 Calculate the distance of any pairs of values in based on (5)–(8);

end for

Generate the input similarity matrix of the mixed dataset using (11);

Set the value preference by the median of similarities;

Input the target number of clusters as ;

while the termination conditions are not met do

 for each running time of AP algorithm do

  if    then

   The value adaptive strategy is defined as , where ;

  else if    then

   The value adaptive strategy is defined as , where ;

  end if

   adaptive strategy is defined by (13);

 end for

end while