How to get DBSCAN to assign the items to the clusters found

Question

NCA 2021-7-7

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/873833-how-to-get-dbscan-to-assign-the-items-to-the-clusters-found

评论： NCA 2021-7-19

I have used DBSCAN on MATLAB R2021 to cluster my data into 3 groups. I want to be able to get DBSCAN to actually assign and name the clusters based on the items. I am not sure how this is done. Can anyone help please?

Thank You

5 个评论
显示 3更早的评论隐藏 3更早的评论

NCA 2021-7-15

在 MATLAB Online 中打开

Hi, thanks for responding to my question. I have discoverd 4 clusters within my data via dbscan clustering. I am not sure how to write a code to get the 4 clusters assigned to the groups. In the groups, I am supposed to have (Low Sulphur, High Sulphur, Very Low Sulphur, etc).

The data I used to generate my clusters was the spectra data from an NIR machine. I also have further information as regards the distribution of the data based on the sulphur content in an excel format elsewhere(this was not used in dbscan analysis)

I tried using a 'for loop' as shown below but the code does not seem to recognise the 'Spectra_50_1.spc' or any other 'Spectra...........' even though they were used without problems in the dbscan analysis ............ after normalizaton and with PCA (principal comp analysis) carried out on it.

for i=1:numel(data)
    
    %if data{i}.Header.Filename == 'Spectra_50_1.spc'
     
       gscatter(x(:,1),x(:,2),idx,'r','.','HighSulfur',10);
       %hold on;
       %else 
        %if data{i}.Header.Filename == 'Spectra_85_1.spc'
         gscatter(x(:,1),x(:,2),idx,'g','.','Lowsulfur',10);
        % end
    %end
%end  

So I have no clue on how to assign these groups based on the sulphur content.

Thanks a lot for your help.

Walter Roberson 2021-7-15

Do not use == to compare character vectors. You can use == to compare string() objects, or to compare a string object and a character vector, but using == to compare character vectors only works if the two are compatible sizes (taking into account implicit expansion)

Character vectors are vectors -- for example 'abc' is the vector ['a', 'b', 'c'] . And when you compare vectors they need to be the same (or compatible) size. You should think of character vectors as being collections of character codes with an internal flag set that says to represent them specially on output. 'A1' is char([65 49]) and should be treated as if it is [65 49] (numeric) rather than being a phrase.

String objects on the other hand, treat entries as collections. ["abc", "hello"] is a 1 x 2 string object that internally is fairly similar to {'abc', 'hello'} in representation, but with added useful behaviour . String objects know that == means to compare the strings rather than the character codes.

To compare character vectors use strcmp() or strcmpi() or isequal() or (in some useful ways) ismember()

NCA 2021-7-19

Thanks a lot for the detailed response,it was really helpful although they only seem to compare two objects while I have many signals within the file to compare to help dbscan algorithm to assign them into the clusters found.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Aditya Patil 2021-7-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/873833-how-to-get-dbscan-to-assign-the-items-to-the-clusters-found#answer_747163

As per my understanding, you have some data that you use for clustering, and some other corresponding data that you want to match to this data. For doing this, you rely on the data.Header.Filename attribute.

The likely issue here is that you need to use strcmp instead of “==”. Another issue might be that Filename has different values than you expect. For example, there might be case difference.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

How to get DBSCAN to assign the items to the clusters found

5 个评论
显示 3更早的评论隐藏 3更早的评论

采纳的回答

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

更多回答（0 个）

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

How to get DBSCAN to assign the items to the clusters found

5 个评论 显示 3更早的评论隐藏 3更早的评论

采纳的回答

0 个评论 显示 -2更早的评论隐藏 -2更早的评论

更多回答（0 个）

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

5 个评论
显示 3更早的评论隐藏 3更早的评论

0 个评论
显示 -2更早的评论隐藏 -2更早的评论