Inclusive Analysis of Incomplete Datasets Using 1Knn Search

Loading...
Thumbnail Image
Files
Date
2017-08-18T10:08:30Z
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Analyzing and processing any dataset is very important for any organization as it helps in making key business decisions of a organization and also increases the profit of any business organization. But these data sets also include incomplete data sets which are often eliminated in the preprocessing techniques. Here, incompleteness refers to the case where data has error or certain information is missing. In this project we focus only on missing data. Missing data exists due to failure of data transmission devices, accidental loss of data or improper storage. Although we can simply perform all the analysis tasks based on complete data sets by removing all the incomplete data, the analysis in incomplete and output is inaccurate. Given a dataset of multi-dimensional objects and a query object, finding k closest objects to the query from the dataset without discarding the incomplete data records (IkNN query) is a fundamental problem in data mining. This concept has its significant role in real time applications like image recognition, location based services, etc. Our objective of this project is to develop and present efficient indexes, pruning techniques and algorithms to support the execution of IkNN queries efficiently. We develop a front end for the data analyst to query the data set using java swing. The java code acts as a client to Rserve which is the backend server in our project. The Rserve accepts the query and executes the search algorithm and provides the k nearest neighbor object of the query to the data analyst. The algorithm uses LαB index, α-pruning and Distance pruning techniques to perform the search efficiently.
Description
Keywords
Inclusive Analysis of Incomplete Datasets Using 1Knn Search, 1NH13IS108, Sujithra K S, Supriya P
Citation
Collections