For the k-means algorithm, the default stop condition is when there is no more shifting of cluster centers in step 5. But as with many other algorithms, k-means algorithms may take lot of time to converge, especially while processing large datasets in a high-dimensional problem space. Instead of waiting for the algorithm to converge, we can also explicitly define the stop condition as follows:

• By specifying the maximum execution time: 
  • Stop condition: t>t_max, where t is the current execution time and t_max is the maximum execution time we have set for the algorithm.
• By specifying the maximum iterations:
  • Stop condition: if m>m_max, where m is the current iteration and m_max is the maximum number of iterations we have set for the algorithm.