Location:
Search - outlier cluster
Search list
Description: 基于密度的聚类算法 DBSCAN java-Density-based clustering algorithm DBSCAN java
Platform: |
Size: 263168 |
Author: 李夏婕 |
Hits:
Description: 横轴和纵轴，可以将样本点表示出来。由于一般第一、二和三主成分包含的样本的信息较多，
经常选其作为横轴和纵轴。当两个样本差异不大时，他们对应的主成分的得分值差异也不会大，
所以可以凭此对样本进行分类。-The quality of the sample
depends on the accuracy of the density estimator chosen, and it needs to decide the kinds of kernel function and the value of its bandwidth and $\alpha$. Besides
there are some drawback about the relation between the value $\alpha$ and the ratio of cluster and outlier. In the future, the extension of
its techniques to data sets of very high dimensionality is possible by taking into account several properties of high-dimensional spaces.
Platform: |
Size: 616448 |
Author: 宋云胜 |
Hits:
Description: Partition index is a measure ofvalidity similar to partition coeGcient, based on using Pj = ci=1 (uij)m as a
measure ofhow well the jth data point has been classi-
2ed. The closer a pixel is to a codebook entry, the closer
Pj
is to one. Ifa pixel becomes an outlier to all cluster centers, the value of Pj approaches 1=cm − 1, which is the minimum value it can have.
Platform: |
Size: 60416 |
Author: dr |
Hits: