Call For Paper Volume:4 Issue:8 Aug'2017 |

LINK RELATIONSHIP BASED DATA PARTITIONING ON HEART DISEASE DIAGNOSIS DATA

Publication Date : 16/01/2016



Author(s) :

Ms. K. Uma, M.Sc., MPhil., Research Scholar, , Mrs. R. Sasiregha., M.Sc., MPhil., ., Assistant Professor,.


Volume/Issue :
Volume 3
,
Issue 1
(01 - 2016)



Abstract :

Abstract   Hidden knowledge discovery is carried out using data mining techniques in a variety of applications. Data partitioning methods are adapted to perform relevant record grouping process. Similarity measures are employed to estimate the similarity measures. Cosine and Euclidean distance measures are .Record link details are also applied to estimate the relationship levels. Link based similarity estimation mechanism is adapted in the data partitioning process. Record links are identified using K Nearest Neighbor (KNN) model. Link relationship is referred as hubness relationship. Link relationship based model is applied for the data partitioning process on high dimensional data environment. Hubness measures are estimated with KNN query results. K Hubs algorithm is applied to define the clusters. Data partitioning operations are carried out using the Hubness-proportional K-means (HPKM) algorithm.   Link relationship based data partitioning methods are employed to perform heart disease analysis.   Heart patient diagnosis data values are grouped using the hubness or link based data partitioning methods. Kernal Mapping Clustering (KMC) scheme and Shared Neighbor Clustering (SNC) scheme are enhanced with link based similarity scheme. The data partitioning process is also improved with automatic cluster count estimation mechanism.  Index Terms: Data clusters, Hubness measure, Heart disease diagnosis, Shared neighbor clustering and Nearest Neighbor search


No. of Downloads :

1


Indexing

Web Design MymensinghPremium WordPress ThemesWeb Development

LINK RELATIONSHIP BASED DATA PARTITIONING ON HEART DISEASE DIAGNOSIS DATA

January 12, 2016