DBSCAN vs OPTICS for Automatic Clustering

algorithm cluster-analysis dbscan optics-algorithm
user3315340 picture user3315340 · Jun 28, 2016 · Viewed 10.1k times · Source

I know that DBSCAN requires two parameters (minPts and Eps). However, I am confused on what parameters are needed for OPTICS because some sources say it requires eps while others say it only requires minPts.

Which algorithm would be the better to use if I were trying to automatically determine the parameter values that would best discard outliers?

Answer

Dylon picture Dylon · Jun 28, 2016

According to the original paper, both minPts and Eps are required. Those sources which say Eps is not required are probably using some method to automatically determine a good value for it. However, Eps is only included to reduce the runtime of the algorithm. It is not required.

Regarding which is best for outlier-removal, there's no better way than to support your decision with numbers: take a dataset and label its outliers, then run both algorithms against it. Use some sort of performance measurement for the clusters (AUC, F-score, etc.) to select the best.

Related questions

Python k-means algorithm

I am looking for Python implementation of k-means algorithm with examples to cluster and cache my database of coordinates.

Read More
Unsupervised clustering with unknown number of clusters

I have a large set of vectors in 3 dimensions. I need to cluster these based on Euclidean distance such that all the vectors in any particular cluster have a Euclidean distance between each other less than a threshold "T". I …

Read More
How to calculate classification error rate

Alright. Now this question is pretty hard. I am going to give you an example. Now the left numbers are my algorithm classification and the right numbers are the original class numbers 177 86 177 86 177 86 177 86 177 86 177 86 177 86 177 86 177 86 177 89 177 89 177 89 177 89 177 89 177 89 177 89 So here my algorithm merged 2 different classes into 1. …

Read More