Evaluating Data Editing and Imputation Methods based on Monte Carlo Technique

• Main Authors: Cheng, Chengyen, 鄭丞晏
• Other Authors: Tsai, Hsiangjung
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/66349977077326286387

碩士 === 國防大學管理學院 === 財務管理學系 === 99 === Most of the survey will have missing data, if the database contains missing values would seriously affect the quality of data analysis, how to properly handle missing values is an important issue. Although a number of imputation methods have been proposed, but not a perfect imputation method can handle different types of missing values well at the same time. The main purpose of this paper would like to obtain not only use of time but also the appropriate data type of imputation methods. To face the database without missing values and then use pseudo-random number selected to make some fields missing. The last is to compare with original value and the value that after imputation. We use three imputation methods- regression imputation, EM imputation, MCMC imputation and compare the imputation method in the data are highly related and low related with the use of time. when dealing with different types of data , the results provide researchers the rule of selecting the appropriate imputation.