The Influence Function and its Application to Data Validation

SYNOPTIC ABSTRACT

Hampel's influence function has been used to detect bivariate observations, which have unusual influence on estimates of correlation. In the validation of energy data bases the identification of such observations may be valuable for the detection of error. When data are used in regression problems, observations which have large effects on multiple correlation coefficients are of interest. Contours of constant influence for the multiple correlation coefficient are given for the case of two regressor variables. For the FPC Form 4 data, the influence function for bivariate correlation is used to detect outliers.

Key Words and Phrases:

• influence function
• outlier detection in multivariate data
• data validation
• bivariate correlation
• multiple correlation
• energy data