Skip to main content
Fig. 2 | Journal of Big Data

Fig. 2

From: S-RASTER: contraction clustering for evolving data streams

Fig. 2

The precision parameter \(\xi\) greatly influences clustering results of RASTER (best viewed in color). This illustration is based on retaining all data points (cf. Fig. 1d). With a precision of \(\xi = 0.90\) (top), all but the rightmost data set are clustered satisfactorily. Reducing the precision to \(\xi = 0.73\) (bottom) improves the results of that data set. It is a matter of debate which value of \(\xi\) led to a better result with the data set in the middle as a good case could be made for either result, depending on whether the goal of the user is to identify dense or sparse clusters. The data sets were taken from a collection of standard data sets for the evaluation of general-purpose clustering algorithms that are part of the machine learning library scikit-learn [27]. A more extensive discussion of these results is provided in a previous paper on RASTER [33]

Back to article page