After understanding the algorithm side, let's see how we can use the Mahout implementation of Streaming K-means.
To run the Streaming K-means, we will first select a dataset. The reference paper Fast and Accurate K-Means for Large Datasets, by M.Schindler, A.Wong, and A.Meyerson mentioned two datasets that they used—BigCross and Census1990.
Here, we will use the Census1990 dataset. This dataset can be downloaded from https://archive.ics.uci.edu/ml/machine-learning-databases/census1990-mld/.
This dataset has 2458285 number of instances and 68 number of attributes. Once you have download this data, you will notice that it is in the CSV format, and we cannot use this directly in Mahout.