Multidimensional Discrete Big Data Clustering Algorithm Based on Dynamic Grid

Traditionally, the data clustering algorithm is lack of comprehensive performance, leading to low clustering purity and long clustering time. In addition, the consistency between the clustering results and the original data distribution is not strong. Therefore, the multidimensional discrete big dat...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Wireless communications and mobile computing Ročník 2022; číslo 1
Hlavní autor: Li, Xiaolei
Médium: Journal Article
Jazyk:angličtina
Vydáno: Oxford Hindawi 2022
John Wiley & Sons, Inc
Témata:
Algorithms
Big Data
Cluster analysis
Clustering
Communication
Data mining
Datasets
Information technology
Methods
Principal components analysis
Purity
ISSN:1530-8669, 1530-8677
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Traditionally, the data clustering algorithm is lack of comprehensive performance, leading to low clustering purity and long clustering time. In addition, the consistency between the clustering results and the original data distribution is not strong. Therefore, the multidimensional discrete big data clustering algorithm based on dynamic grid was put forward. Firstly, multidimensional discrete big data was processed in advance. The principal component analysis was used to reduce the dimension of data. The concept of entropy was introduced to divide the key attributes and noncritical attributes, so as to extract the key attributes. According to the results of data preprocessing, the dynamic grid was partitioned. According to the results, OptiGrid in data clustering algorithm was used to achieve the data clustering. The experimental results show that the clustering purity of this algorithm is between 95% and 100%, which is significantly higher than the traditional algorithm. Therefore, the multidimensional discrete big data clustering algorithm based on dynamic grid has better comprehensive performance, closer clustering shape to the original data distribution, higher clustering purity, and faster execution efficiency.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1530-8669
1530-8677
DOI:10.1155/2022/4663816