This paper explores the associations between personality traits and performance indicators. CCA (Canonical Correlation Analysis) is applied to analyze and optimize the correlations between the two groups of original data, find out the maximum canonical correlation coefficient, convert raw variables into canonical variables. Using sample data, the authors make a detailed explanation for modeling process, showing the dimension reduction effects. The comprehensive potential index of potential data and comprehensive performance evaluation index of performance data are developed, significant linear correlation between them is found, and then the regression equations are established with the two groups of raw data. By vertical and horizontal analysis of the predictive model, we get some important statistics correlations between personal factors and performance indicators. The results indicate that the method in this paper can utilize the advantages of CCA on dimension reduction and optimization for high dimension random vectors, decrease the complexity of the algorithm.
[1]
John Shawe-Taylor,et al.
Convergence analysis of kernel Canonical Correlation Analysis: theory and practice
,
2008,
Machine Learning.
[2]
Jiliu Zhou,et al.
Improved Canonical Correlation Analysis and Its Applications in Image Recognition
,
2010
.
[3]
Hai-bin Shen,et al.
Regularized canonical correlation analysis with unlabeled data
,
2009
.
[4]
Handan Ankarali,et al.
An application of nonlinear canonical correlation analysis on medical data
,
2010,
Turkish Journal of Medical Sciences.
[5]
Saman K. Halgamuge,et al.
Empirical relationships between some manufacturing practices and performance
,
2008
.
[6]
Toru Ogura,et al.
A variable selection method in principal canonical correlation analysis
,
2010,
Comput. Stat. Data Anal..