Hi,
I have a training dataset with many identical datapoints. I use batch-mode to train the model. Thereafter, when I insert a point that is identical to a subset of points in the training dataset, the point will displace all its existing copies. This results in a high (co)displacement-score for this point, even though the point is very common.
Update: setting the tolerance to 0 when inserting a point did the trick.
Hi, I have a training dataset with many identical datapoints. I use batch-mode to train the model. Thereafter, when I insert a point that is identical to a subset of points in the training dataset, the point will displace all its existing copies. This results in a high (co)displacement-score for this point, even though the point is very common.
Update: setting the tolerance to 0 when inserting a point did the trick.