If the two data sets are from the same group, you can look at cluster.stats (available in R). This function will give you some stats on clusters comparison.

Reference:

https://www.rdocumentation.org/packages/fpc/versions/2.1-11/topics/cluster.stats

Hi,

If I understand you correctly, then here is my suggestion:

You can merge the 2 datasets and run a clusterer method on the resulting data. Then, explore the instances of each cluster to see which data they originally belong to. Another option is to cluster then label your dataset 2l using clusters defined by dataset 1. In other words, try to label the first dataset and then cluster it using K-means algorithm, for instance. After that, add the instances of dataset 2 (one-by-one) into dataset 1 and run K-means at each runtime. This helps through assigning the new instance(s) to certain cluster. Then, you run same clustering algorithm on dataset 2 alone. Now for each point you know in which of 3 clusters it was put based on the dataset 1, and in which one - based on the dataset 2. You can calculate the share of this overlap, which is already telling.

Please note that if you don't have labels then there is no actual/predicted labels for the utilized instances.

HTH.

Dr. Samer Sarsam

Thank you for your reply

The two data sets are similar in the number of objects and type of features, but they are different because they are from two different places and different years, without classes. Actually, the number of records is large just I want to understand and compute the similarity between the two clusters result (outcomes).

You could use something like the Jaccard index (https://en.wikipedia.org/wiki/Jaccard_index) or the Normalized Mutual Information (https://course.ccs.neu.edu/cs6140sp15/7_locality_cluster/Assignment-6/NMI.pdf) to measure how similar two clusters are. Although i am not sure how to applied them to a fuzzy cluster.

Normally there is no ground truth in clustering. You can use measures for the density of the clusters or the well separateness such as the Dunn or Davies-Bouldin index.

Use evaluation metrics.

A value close to 1 means they are very similar, close to 0 is random, and much less than 0 means each cluster of one is "evenly" distributed over all clusters of the other.

http://scikit-learn.org/stable/modules/generated/sklearn.metrics.adjusted_rand_score.html

A combination evaluation between distance and entropy

Euclidean distance

I recommend t-test if sample size is less than 30 and z-test its more than 30.