Mockiato is a visualisation tool for MOCK. It was created by me, Joe Forshaw, as part of my final year project of my Computer Science degree at the University of Manchester.
MOCK stands for Multiobjective Clustering with K-determination. It is clustering algorithmn created by Joshua Knowles and Julia Handl at the University of Manchester.
Mockiato was created to provide users a more accessible way of analysing the output of MOCK.
Multiobjective clustering is a field of cluster analysis where several objectives (i.e. measurements of cluster quality) are optimised in an attempt to find the best overall solution.
MOCK uses two complimentary objectives to cluster data:
-
Overall Deviation - is how compact clusters are. It’s equal to the overall summed distances between every data point and their cluster centre.
-
Connectivity - is the proportion of neighbouring data points that belong in the same cluster. A low connectivity means data points are closest to other points in the same cluster.
By analysing a range of solutions from a pareto front using overall deviation and connectivity, we can choose a solution that achieves the best compromise of both objectives.