Within or between clusters sum of squares?

[ldeoto]

Hello,

I am new to pygeoda. I am testing the SKATER algorithm and I find out that the summary stats do not coincide with those returned by the software. It appears to me that under "Total within-cluster sum of squares" the displayed value corresponds to the between-cluster sum of squares. The actual total within-cluster sum of squares does not show.
Would you confirm this?
Thanks for your help

[lixun910]

I am not sure if I understand your question. Does the output of the skater function look like this?:

>>> skater_clusters = pygeoda.skater(4, queen_w, data)
>>> skater_clusters
{'Total sum of squares': 504.0000000000001,
'Within-cluster sum of squares': [57.890768263715266,
59.95241669262987,
28.725706194374844,
69.3802999471999,
62.30781060793979,
66.65808666485573],
'Total within-cluster sum of squares': 159.0849116292847,
'The ratio of between to total sum of squares': 0.3156446659311204,
'Clusters': (3, 2, 3, 1, 1, 1, 2, 1,...)
}

[ldeoto]

Hi @lixun910
Yes, it looks like that. So if you see: 'Total within-cluster sum of squares' should be the sum of all the 'Within-cluster sum of squares', that would be 344.91...
The value 159.08... corresponds to the between-cluster sum of squares (i.e. Total sum of squares minus Total within-cluster sum of squares). That's why 'The ratio of between to total sum of squares' is 0.31... (i.e 159.08 / 504)
So what is now called 'Total within-cluster sum of squares' is the Total between-cluster sum of squares. This should be adjusted.

[lixun910]

I see. Yes, you are right: 159.08 should be the Total between-cluster sum of squares. I will fix it. Thanks, @ldeoto !

[ldeoto]

Great! Thanks @lixun910 ! 👍