1: \begin{abstract}

2: Reduced $k$-means clustering is a method for clustering objects in 

3: a low-dimensional subspace.

4: The advantage of this method is that

5: both clustering of objects and low-dimensional subspace reflecting 

6: the cluster structure are simultaneously obtained.

7: In this paper,

8: the relationship between conventional $k$-means clustering 

9: and reduced $k$-means clustering is discussed.

10: Conditions ensuring almost sure convergence of 

11: the estimator of reduced $k$-means clustering as unboundedly increasing sample size have been presented.

12: % In these conditions, 

13: % there are no existence conditions of parameters that 

14: % achive the infimum value of the population objective function of reduced $k$-means clusterings.

15: %The result is proved for a more general model including the usual $k$-means clustering and reduced $k$-means clustering.

16: The results for a more general model considering conventional $k$-means clustering and reduced $k$-means clustering

17: are provided in this paper.

18: Moreover, 

19: a new criterion and its consistent estimator are proposed to determine the optimal dimension number of a subspace, given the number of clusters.

20: \end{abstract}