Density peak clustering is an effective and novel clustering algorithm

it is concerned as its superiority of finding arbitrary shape of clusters and number of clusters. However

this algorithm is required to measure the density and distance between any pair of objects. This limits the practicability of this algorithm when clustering high-volume and high-dimensional data set. In order to improve efficiency and scalability

we propose a distributed density peak clustering algorithm based on z-value

and DP-z. It utilizes z-values to map points in multidimensional space into one dimension

and then splits the data set into several partitions according to the z-values of points. In order to get the correct result

we make use of the character of points' z-values to filter the data object while exchanging data among groups

which reduces a huge amount of useless distance measurement cost and data shuffle cost. Then we compute the density and distance value in parallel. Finally

we test the DP-z algorithm based on the cloud computing platform

the experiments show that DP-z can achieve higher performance at speed without reducing the accuracy.