Timeline for Clustering unrelated (with no correlation) data
Current License: CC BY-SA 3.0
20 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 20, 2016 at 13:53 | history | post merged (destination) | |||
| Oct 20, 2016 at 12:40 | comment | added | tripleee | For a somewhat more intelligent approach, maybe use several clustering methods and collect the outliers which are not accepted into a cluster by any of those methods. | |
| Oct 20, 2016 at 12:39 | comment | added | tripleee | The inverse of clustering is identifying outliers. You collect the samples which do not fall into a cluster by any criteria into a separate group. Calling this group a "cluster" is misleading but this seems to be what you are asking. | |
| S Oct 18, 2016 at 13:15 | history | suggested | Manuel | CC BY-SA 3.0 |
Progress datails
|
| Oct 18, 2016 at 12:56 | comment | added | EngrStudent | @Manuel - the k-means is a blunt instrument. It assumes the variance of the various clusters are uniform - the real world doesn't actually do that. "All models are wrong ..." but some are more wrong than others and they have to be evaluated before they inform next steps. A gaussian mixture model (GMM) is a decent higher-tech check for your k-means. If the minimum AICc happens at the same cluster count AND if the variances of the GMM components are all very close to each other, then you might go forward with k-means. Otherwise, don't be married to 9 clusters. | |
| Oct 18, 2016 at 12:52 | review | Suggested edits | |||
| S Oct 18, 2016 at 13:15 | |||||
| Oct 16, 2016 at 20:17 | comment | added | Pere | @Manuel: I don't have references, but I made an example. | |
| Oct 16, 2016 at 20:16 | answer | added | Pere | timeline score: 1 | |
| S Oct 14, 2016 at 0:11 | history | edited | gung - Reinstate Monica | CC BY-SA 3.0 |
Provide more details
|
| S Oct 14, 2016 at 0:11 | history | suggested | Manuel | CC BY-SA 3.0 |
Provide more details
|
| Oct 14, 2016 at 0:08 | comment | added | gung - Reinstate Monica | Please register &/or merge your accounts (you can find information on how to do this in the My Account section of our help center), then you will be able to edit & comment on your own question. | |
| Oct 13, 2016 at 23:56 | comment | added | GeoMatt22 | The description is reminiscent of graph cliques vs. independent sets. So the suggestion of @Pere is similar to "inverting the weights" to form a complement graph. Perhaps a similarity based approach would be better than a distance based approach? (e.g. something like this?) | |
| Oct 13, 2016 at 23:56 | review | Suggested edits | |||
| S Oct 14, 2016 at 0:11 | |||||
| Oct 13, 2016 at 23:46 | comment | added | Danica | Are you really trying to find a single, diverse group? Do you want as many diverse groups as possible? | |
| Oct 13, 2016 at 23:36 | comment | added | Manuel | Thank you for your suggestion. To be honest I only know how to perform the inverse of functions! Can you please provide how I can perform in my correlation matrix? or indicate a reference/toolbox? | |
| Oct 13, 2016 at 19:01 | review | First posts | |||
| Oct 13, 2016 at 19:25 | |||||
| Oct 13, 2016 at 18:53 | comment | added | Pere | You just need to define a "distance" that is inverse of usual distance. It won't fit the definition of distance in algebra or topology but for most clustering methods that shouldn't be a problem. Anyway, I would expect rather strange groups and some instability among groups. | |
| Oct 13, 2016 at 18:53 | comment | added | whuber♦ | Because this is such an unusual request, could you please provide more specifics about what this grouping is intended to achieve? It will come down to this question (abstractly): given two candidate groupings, exactly how would you decide which one is a better solution than the other? | |
| Oct 13, 2016 at 18:49 | history | asked | Manuel | CC BY-SA 3.0 | |
| Oct 13, 2016 at 18:00 | comment | added | agenis | well, you said it yourself, you want to perform the "inverse", so you can try to take the inverse (algebric) of your correlation matrix |