I have a set of real data containing labels. However, in some biclustering algorithms (e.g., Cheng and Church Algorithm), originally authors applied gene expression data like Yeast without having labels. My question is that what are appropriate evaluation measures in this scenario?

  • $\begingroup$ Welcome to the site. I understand machine learning (ML) but the question is not clear. The key Issue is the use of links. We will not click two external links to understand the question, so it needs a summary of the intended algorithm is needed within the question. Biclustering sounds to me to be unsupervised learning and that would explain the difference. More generally, explaining what data you have and what the labels are is usually helpful. $\endgroup$
    – M__
    Aug 3, 2023 at 14:56
  • 1
    $\begingroup$ Biclustering or co-clustering is an unsupervised learning task with which both rows and columns in a data matrix can be placed in concurrent clusters whereas non-overlapping clusters are generated in clustering algorithms. Data set - a resource of regulatory elements (REMs) contains features like Gene ID, REMID, Genomic location, Predicted function (acts as a label having values such as activation, repression), Model score (normalized absolute binary logarithm of the P-value). $\endgroup$
    – ShokofehVS
    Aug 4, 2023 at 11:17

1 Answer 1


Briefly evaluation measures, which I presume means "features", are determined using supervised learning using a technique called "feature selection". Unsupervised learning generally is used to identify categories for a supervised learning algorithm. The orthodoxy has changed.

"Unsupervised learning feature selection"

  1. To identify suitable features using unsupervised learning a priori then genome location which must have a major outcome, thereafter it would be a case of deleting a given feature and re-evaluating the clusters. If the clusters disintegrate then that is a key feature (not very cool but would work for datasets with limited numbers of features).
  2. Weighted unsupervised learning is popular, this is basically "feature selection", i.e. which features contribute to the clustering. This is basically the formal version of point 1. Personally I have had little success with this approach, i.e. the results were not informative, but it does work. The results are not discriminatory between different conditions, which is what I was seeking.

Regarding labels, labels are not used in unsupervised learning approach - except weighted unsupervised learning - hence their presence or absence usually makes little difference. The only "label" used is the sample type, i.e. the "id" and not the "feature". Without intrinsically understanding the experiment it is difficult to forward an accurate a priori assessment.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.