東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Data mining techniques for handling ...

Siripitayananon, Punnee.

Linked to FindBook

Google Book

Amazon

博客來

Data mining techniques for handling a missing data problem.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Data mining techniques for handling a missing data problem./
Author:	Siripitayananon, Punnee.
Description:	159 p.
Notes:	Source: Dissertation Abstracts International, Volume: 63-12, Section: B, page: 5941.
Contained By:	Dissertation Abstracts International63-12B.
Subject:	Computer Science. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3075151
ISBN:	049395791X

Data mining techniques for handling a missing data problem.
Siripitayananon, Punnee.

Data mining techniques for handling a missing data problem. - 159 p.

Source: Dissertation Abstracts International, Volume: 63-12, Section: B, page: 5941.

Thesis (Ph.D.)--The University of Alabama, 2002.

Today, faster and cheaper storage technology allows us to store data in tera-byte units and provides easy access to those databases. In the conventional programming algorithm, we usually assume that all input data are correct and complete. A bug-free computer program should be able to produce the expected output correctly. However, it is not uncommon for observations to be missing. Missing data can cause considerably wrong or distorted output in all processes that use these data for determination. Therefore, it is crucial to verify all input data before feeding it into any particular computer program.

ISBN: 049395791XSubjects--Topical Terms:

626642
Computer Science.

Data mining techniques for handling a missing data problem.
LDR:03139nmm 2200313 4500 001 1838384
005 20050526083748.5
008 130614s2002 eng d
020 $a 049395791X
035 $a (UnM)AAI3075151
035 $a AAI3075151
040 $a UnM $c UnM
100 1 $a Siripitayananon, Punnee. $3 1926802
245 1 0 $a Data mining techniques for handling a missing data problem.
300 $a 159 p.
500 $a Source: Dissertation Abstracts International, Volume: 63-12, Section: B, page: 5941.
500 $a Chairperson: Hui-Chuan Chen.
502 $a Thesis (Ph.D.)--The University of Alabama, 2002.
520 $a Today, faster and cheaper storage technology allows us to store data in tera-byte units and provides easy access to those databases. In the conventional programming algorithm, we usually assume that all input data are correct and complete. A bug-free computer program should be able to produce the expected output correctly. However, it is not uncommon for observations to be missing. Missing data can cause considerably wrong or distorted output in all processes that use these data for determination. Therefore, it is crucial to verify all input data before feeding it into any particular computer program.
520 $a There are many reasons why some data in a data set are missing. In the case of data acquired from some automatic instruments, missing data occur periodically when instruments fail. In many of these situations, missing data cannot be re-collected or reproduced especially if they are time series data. Therefore, it is important that efficient methods for handling missing data be available to minimize the loss, particularly where a complete data set is desired.
520 $a The purpose of this dissertation is to apply data mining techniques as well as time series analysis for estimating missing data that occur lengthy and consecutive sections. This dissertation primarily studies the cases of multiple time series that have missing data in one series whereas other series are available. Several traditional data mining approaches were modified to enhance the accuracy for estimating missing data. Several new ideas are presented when modifying a variety of methods. These new ideas use z-score conversion, time lag analysis, altitude adjustment, best correlations, transition matrix of differencing, synthesized time series, nearest neighbor, and a new distance function for the integration purpose.
520 $a The proposed approaches demonstrate their abilities for the weather data (wind speed and air temperature, for the years 1996 and 1999), and for the current speed data for the year 2000. The errors of estimating by all proposed approaches are very small for all three applications. The performances of the proposed approaches are comparable to each other but much more favorable than those of the traditional methods.
590 $a School code: 0004.
650 4 $a Computer Science. $3 626642
650 4 $a Hydrology. $3 545716
690 $a 0984
690 $a 0388
710 2 0 $a The University of Alabama. $3 1019361
773 0 $t Dissertation Abstracts International $g 63-12B.
790 1 0 $a Chen, Hui-Chuan, $e advisor
790 $a 0004
791 $a Ph.D.
792 $a 2002
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3075151