Introduction to Data Clustering Seminar Topic:
It means collection of useful data into the groups. These groups are called clusters. It means classification of similar type of data. This reduces the complexity of the data and also reduces the number of bits required in the computer science. Clustering also includes pattern representation, extraction and selection. Clustering also depends on the data representation and if the data is good clusters will be compact and isolated. Sometimes the number of clusters is defined automatically; it is also the problem in clustering. There are many types of clustering:
Hierarchical clustering: this method is implemented on the 2-d types of data. They are mostly used in the pattern clustering. In which distance between the two clusters are minimum distance and rest all the pattern are drawn from these two clusters. In this each pattern we draw is own cluster.
Partition clustering: it generated single cluster instead of structure. They are best for the large data sets. They also work on isolated and compact clusters type of errors. They are easy to implement and less time consuming.
Nearest neighbor clustering: in this nearest node represent the basis of clustering. Each unlabeled pattern is assigned its nearest pattern.
Fuzzy clustering: in this every pattern belongs to one and only one cluster or belongs to the member function.
Applications: they are used in the information retrieval systems like searching books from library. If you want to search a book from the library and books are classified according to the group ad use then it will be easy to select. They are also used in the pattern recognition techniques using ACM CR classification and image segmentation. They are also used for data mining for transactions and relational databases with well-defined fields and keys.
Download Data Clustering Seminar Report.