A new approximation for clustering of directional data is proposed.
The proposed method is simple, fast and stable when compared to other methods.
The proposed method can be used for N-dimensional directional data.
Tradition validation index methods can be used with proposed method.