How can i use DBSCAN clustering algorithm for a dataset having multiple attributes?

Question

i am working on a project in which i am using a dataset -https://www.kaggle.com/aljarah/xAPI-Edu-Data . I want to do clustering of students (each student represented as index), based upon various attributes of the dataset like raised hands, visited resources, announcements viewed, etc. Please suggest how can i implement this using DBSCAN, if not please propose some technique through which i can do it. I am a newbie in this field of data science.

Thanks

i tried studying gmm and dbscan.

i want to do clustering on a dataset.

score 0 · Answer 1 · answered Mar 31 '19 at 10:44

0

Any standard implementation of DBSCAN will support multiple attributes.

Mostly it will depend on your decision how to measure similarity, when attributes have very different type. Euclidean distance will likely not make sense. But there is no "correct" way of doing these, it's your decision on how to model the data. On this data set, it will be rather arbitrary, unfortunately, because these attributes have no natural scale.

answered Mar 31 '19 at 10:44

Has QUIT--Anony-Mousse

76,138
12
138
194

hi thanks for the reply, ca you suggest me some implementation of dbscan with multiple attributes ? – aditya verma Apr 01 '19 at 11:48
I doubt you can find a good implementation that only allows a single attribute. Because for univariate data, there are much better approaches. – Has QUIT--Anony-Mousse Apr 01 '19 at 15:13
can you please name some better approaches? – aditya verma Apr 02 '19 at 04:02
can i take an average of the values of attributes that i am judging on, and then doing clustering using dbscan ? – aditya verma Apr 02 '19 at 04:03

How can i use DBSCAN clustering algorithm for a dataset having multiple attributes?

1 Answers1