August 2019
Intermediate to advanced
342 pages
9h 35m
English
One of the recurring problems with clustering algorithms is the evaluation of the results.
While in the case of supervised algorithms, by already knowing the classification labels, we are able to evaluate the results obtained by the algorithm simply by counting out the number of samples incorrectly classified and comparing them with those correctly classified. In the case of unsupervised algorithms, the evaluation of results is less intuitive.
Not having the classification labels available beforehand, we will have to evaluate the results by analyzing the behavior of the algorithm itself, only considering the clustering process as successful if the samples classified in the same cluster ...
Read now
Unlock full access