I have a dataset of houses like this:

HouseID  Latitude Longitude PriceIndex
  1          1.4     103.120    1.21
  2          1.42    103.112    2.01 

I want to find houses which are similar to each other both on the basis of their position as well as their price index.[Also would need to Rank in order of similarities, given one house] I tried using hclust package in R and was able to extract 9 classes. However the groups don't seem to have any interpretable similarities (for example the points are spread all across the city etc). I haven't done clustering based projects before so any help in the right direction will be helpful. Thanks!

Edit: Removing the price index column from the clustering data-set actually clusters spatially. But adding the price shows only price-based clustering

  • $\begingroup$ Did you try doing k-means clustering? $\endgroup$ – Dawny33 Dec 1 '15 at 6:39
  • $\begingroup$ @Dawny33 :yes, similar results. $\endgroup$ – UD1989 Dec 1 '15 at 6:53

Check the ranges of your dimensions and consider scaling if you see a large difference.

I would interpret your described behaviour due to much larger range if the index that the two other dimensions. See also the question.

| improve this answer | |
  • $\begingroup$ Yes! Scaling solved the problem. I had scaled the variables individually but not with respect to each other -how stupid. Thanks a lot! $\endgroup$ – UD1989 Dec 1 '15 at 9:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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