Contents
Can we use binary variables for clustering?
Yes, it is unlikely that binary data can be clustered satisfactorily. For binary data, the Euclidean distance measure used by K-Means reduces to counting the number of variables on which two cases disagree.
How do you use Gower distance?
Briefly, to compute the Gower distance between two items you compare each element and compute a term. If the element is numeric, the term is the absolute value of the difference divided by the range. If the element is non-numeric the term is 1 if the elements are different or the term is 0 if the elements are the same.
Is it possible to cluster binary data with k-means?
Yes, it is unlikely that binary data can be clustered satisfactorily. To see why, consider what happens as the K-Means algorithm processes cases. For binary data, the Euclidean distance measure used by K-Means reduces to counting the number of variables on which two cases disagree.
Is it possible to cluster binary valued data?
But I have heard that it is inappropriate to cluster binary-valued data. Is this true? Yes, it is unlikely that binary data can be clustered satisfactorily. To see why, consider what happens as the K-Means algorithm processes cases.
Which is an alternative strategy for clustering binary data?
An alternative strategy which is sometimes employed is to run factor analysis or principal component analysis on the binary variables, saving the factor or component scores as new variables and clustering the cases on the basis of those scores. Thus, the data being clustered are no longer binary.
When to use hierarchical clustering with cosine distance?
But it is more important to pick a binary metric that matches you requirements. If you have an appropriate similarity metric, it would also be helpful to visualize the data with MDS methods (or non-linear dimensionality reduction) in 2D or 3D space. I used hierarchical clustering with cosine distance for a similar problem and it worked well.