How to estimate the density threshold separately for each dimension.

1 次查看（过去 30 天）

Silpa K 2020-1-7

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/499175-how-to-estimate-the-density-threshold-separately-for-each-dimension

评论： Silpa K 2020-1-16

Ecolixl.xlsx

I need to find the density threshold separately for each dimension.

In this step,objects in each of the D attributes are arranged in increasing order.A closely packed 5 elements region in it is identified and in case of tie,it is resolved randomly.The window size is set to 5 elements (subspace clusters having less than 5 objects are assumed to be non significant).The maximum seperation between two consecutive elements in the group is rounded to next decimal point and is recorded as density threshold.

How can I do that please help me. The dataset is attached here.

11 个评论
显示 9更早的评论隐藏 9更早的评论

Silpa K 2020-1-8

编辑：Silpa K 2020-1-8

在 MATLAB Online 中打开

I need to estimate the density threshold seperately for each dimension,using a window.The maximum seperation between two consecutive elements in the group is rounded to next decimal point and is recorded as the density threshold for that dimension.

First finding the density of each dimension after that I need to form 1-D cluster using supspace clusters for each dimension.

data=xlsread('Ecolixl.xlsx'); 
        figure(); 
       plot(data(:,1),data(:,2), '-ob') 
       axis equal
       hold on
       windowSize = [6,20];
       ouCount = inCount; 
       onCount = 0; 
       wh = gobjects(1); 
        th = text(min(xlim()),max(ylim()),'Count: 0', 'VerticalAlignment', 'top','FontSize',14);
        for i = 1:size(data,1)
            windCoord = data(i,:) + windowSize/2.*[1,1;-1,-1]; 
            [in,on] = inpolygon(data(:,1),data(:,2),windCoord([2,1,1,2,2]),windCoord([4,4,3,3,4]));
             inCount(i) = sum(in); 
              onCount(i) = sum(on); 
              delete(wh)
              wh = plot(windCoord([2,1,1,2,2]),windCoord([4,4,3,3,4]), 'r-');
              th.String = sprintf('Count: %d',inCount(i)); 
              drawnow
              pause(.1)
                    
             
        end

I used the above code form that site.

Then Iam getting error

Matrix dimensions must agree.

Error in test1 (line 22)

windCoord = data(i,:) + windowSize/2.*[1,1;-1,-1]; %[UpperLeft, UpperRight; LowerLeft, LowerRight]

coordinates .

Image Analyst 2020-1-16

Well that's what it's famed for. So, over the past 7 days, did you actually try it? If not, why not. Did you at least try any examples in the help? If not, why not? And please don't say something like "I'm just a beginner so I can't run examples from the help."

Silpa K 2020-1-16

Yes sir I tried. Iam not a beginner but also Iam not an expert.Over the past days I studied and understand the problem. I also tried lots of examples using dbscan() and distance measures. But I didn't get the accurate answer. I need help.Thank you.

请先登录，再进行评论。

请先登录，再回答此问题。