Investigating the Relative Importance of Individual Variables and Variable Subsets in Discriminant Analysis

1973 ◽  
Vol 2 (3) ◽  
pp. 205-219 ◽  
Author(s):  
Robert Eisenbeis ◽  
Gary Gilbert ◽  
Robert Avery
Keyword(s):  
Discriminant Analysis ◽  
Relative Importance ◽  
Individual Variables

AN EVALUATION OF THE USE OF ELYTRA AND BODIES IN X-RAY ENERGY-DISPERSIVE SPECTROSCOPIC STUDIES OF THE RED TURNIP BEETLE, ENTOMOSCELIS AMERICANA (COLEOPTERA: CHRYSOMELIDAE)

1980 ◽  
Vol 112 (6) ◽  
pp. 609-614 ◽  
Author(s):  
W. J. Turnock ◽  
G. H. Gerber ◽  
D. U. Sabourin
Keyword(s):  
Discriminant Analysis ◽  
Energy Dispersive Spectroscopy ◽  
Spectroscopic Studies ◽  
Energy Dispersive ◽  
Spectroscopy Technique ◽  
X Ray ◽  
Insect Dispersal ◽  
Canonical Variables ◽  
The Mean ◽  
Individual Variables

AbstractSamples of the bodies and elytra of Entomoscelis americana Brown were analyzed separately by X-ray energy-dispersive spectroscopy. Discriminant analysis revealed that the chemoprints of the bodies of newly-emerged beetles (1 wk old) were distinct from those of post-aestivation beetles (9- to 10-wk old). However, the chemoprints of the elytra of newly-emerged and post-aestivation beetles were not as different as those of the bodies as demonstrated by the overlap in the plot of the first two canonical variables. The variances of the mean difference of individual variables between newly-emerged and post-aestivation groups generally were smaller for elytra than for bodies. This suggests that the chemoprints of the elytra of E. americana are more stable than those of the bodies and consequently the elytra should be more suitable than the bodies in insect dispersal studies utilizing the X-ray energy-dispersive spectroscopy technique.

scholarly journals An Empirical Study of Greedy Kernel Fisher Discriminants

2015 ◽  
Vol 2015 ◽  
pp. 1-12
Author(s):  
Tom Diethe
Keyword(s):  
Discriminant Analysis ◽  
Matching Pursuit ◽  
Support Vector ◽  
Relative Importance ◽  
Fisher Criterion ◽  
Fisher Discriminant Analysis ◽  
Original Algorithm ◽  
Fisher Discriminant ◽  
Kernel Fisher Discriminant Analysis ◽  
Vector Machines

A sparse version of Kernel Fisher Discriminant Analysis using an approach based on Matching Pursuit (MPKFDA) has been shown to be competitive with Kernel Fisher Discriminant Analysis and the Support Vector Machines on publicly available datasets, with additional experiments showing that MPKFDA on average outperforms these algorithms in extremely high dimensional settings. In (nearly) all cases, the resulting classifier was sparser than the Support Vector Machine. Natural questions that arise are what is the relative importance of the use of the Fisher criterion for selecting bases and the deflation step? Can we speed the algorithm up without degrading performance? Here we analyse the algorithm in more detail, providing alternatives to the optimisation criterion and the deflation procedure of the algorithm, and also propose a stagewise version. We demonstrate empirically that these alternatives can provide considerable improvements in the computational complexity, whilst maintaining the performance of the original algorithm (and in some cases improving it).

scholarly journals Bullying Victimization among Mexican Adolescents: Psychosocial Differences from an Ecological Approach

2020 ◽  
Vol 17 (13) ◽  
pp. 4831
Author(s):  
Silvana Mabel Nuñez-Fadda ◽  
Remberto Castro-Castañeda ◽  
Esperanza Vargas-Jiménez ◽  
Gonzalo Musitu-Ochoa ◽  
Juan Evaristo Callejas-Jerónimo
Keyword(s):  
Logistic Regression ◽  
Regression Analysis ◽  
Psychological Distress ◽  
Discriminant Analysis ◽  
Logistic Regression Analysis ◽  
Positive Attitude ◽  
Bullying Victimization ◽  
Open Communication ◽  
Individual Variables ◽  
Psychosocial Differences

This transversal study over a random representative sample of 1687 Mexican students attending public and private secondary schools (54% girls, 12–17 years old, M = 13.65. DT = 1.14) aimed to analyze psychosocial differences between victims and non-victims of bullying from the bioecological model. It included individual variables (ontosystem), familiar, community, and scholar factors (microsystem), and gender (macrosystem) to perform a multivariate discriminant analysis and a logistic regression analysis. The discriminant analysis found that psychological distress, offensive communication with mother and father, and a positive attitude toward social norms transgression characterized the high victimization cluster. For the non-victims, the discriminant variables were community implication, positive attitude toward institutional authority, and open communication with the mother. These variables allowed for correctly predicting membership in 76% of the cases. Logistic regression analysis found that psychological distress, offensive communication with the father, and being a boy increased the probability of high victimization, while a positive attitude toward authority, open communication with the mother, and being a girl decrease this probability. These results highlight the importance of open and offensive communication between adolescents and their parents on psychological distress, attitude toward authority, community implication, and bullying victimization.

scholarly journals Die voorspelling van kredietrisiko met behulp van biografiese veranderlikes: Is dit die antwoord?

1983 ◽  
Vol 14 (4) ◽  
pp. 166-171
Author(s):  
P. L.S. Ackermann ◽  
W. P. Jansen van Rensburg
Keyword(s):  
Regression Analysis ◽  
Discriminant Analysis ◽  
Credit Risk ◽  
Representative Sample ◽  
Stepwise Regression ◽  
Stepwise Regression Analysis ◽  
Stepwise Discriminant Analysis ◽  
Relative Importance ◽  
Multiple Stepwise Regression Analysis

The prediction of credit risk by means of biographic variables: Is this the answer? The objective of this study is to identify specific biographical variables, to quantity them and to investigate their relative importance in the prediction of credit risk. A representative sample of 250 bad credit risk clients and 250 good credit risk clients is used in the study. A multiple stepwise regression analysis and multiple stepwise discriminant analysis were carried out. Nine biographical variables were identified which explain approximately 16% of the variance of credit risk.

Perceived Role Characteristics and Executive Satisfaction in Voluntary Sport Associations

1987 ◽  
Vol 9 (4) ◽  
pp. 376-384 ◽  
Author(s):  
Geneviève Rail
Keyword(s):  
Regression Analysis ◽  
Demographic Data ◽  
Individual Characteristics ◽  
Role Satisfaction ◽  
Relative Importance ◽  
Amateur Sport ◽  
Individual Variables ◽  
Positive Correlations ◽  
Satisfaction Measures ◽  
The Relationship

This study tested the relationship between perceived role characteristics and role satisfaction among sport executives. It also investigated the relative importance of role characteristics and individual variables in the prediction of role satisfaction. Measures of perceived role characteristics and role satisfaction were obtained through content analysis of interviews with 60 executives involved in Quebec amateur sport federations. Demographic data were gathered by questionnaire. Results indicated positive correlations between perceived role characteristics and role satisfaction. As demonstrated by multiple regression analysis, the selected individual characteristics (age and marital status) were not predictive of role satisfaction. Use of competence, autonomy, role significance, and recognition were found to be the four major determinants of role satisfaction within the voluntary sport associations.

Prediction of Consumer Innovators: Application of Multiple Discriminant Analysis

1968 ◽  
Vol 5 (1) ◽  
pp. 64-69 ◽  
Author(s):  
Thomas S. Robertson ◽  
James N. Kennedy
Keyword(s):  
Discriminant Analysis ◽  
Innovation Diffusion ◽  
Predictive Value ◽  
Social System ◽  
Predictive Ability ◽  
Relative Importance ◽  
Multiple Discriminant Analysis ◽  
Socioeconomic Characteristics ◽  
Analysis Techniques

Socioeconomic characteristics of consumer applicance innovators and non-innovators within a defined social system are assessed. Such characteristics are derived from the innovation-diffusion literature and represent variables of highest predictive ability in previous research. The relative importance of each characteristic and the predictive value of the set of characteristics are measured with multiple discriminant analysis techniques.

Opening the Black Box of Social Capital Formation

2014 ◽  
Vol 108 (1) ◽  
pp. 121-143 ◽  
Author(s):  
PATRICIO VALDIVIESO ◽  
BENJAMÍN VILLENA-ROLDÁN
Keyword(s):  
Social Capital ◽  
Peer Effects ◽  
Interpersonal Trust ◽  
Choice Model ◽  
Black Box ◽  
Relative Importance ◽  
Individual Level ◽  
Heterogeneous Effects ◽  
Rational Choice Model ◽  
Individual Variables

This paper introduces a rational choice model for multiple kinds of participation to empirically investigate several theoretical determinants of social capital (SC) formation. The framework is rich enough to investigate the importance of individual variables, social/peer effects, endogenous trust, political-institutional, and inequality factors as sources of participation. We show that the aforementioned contextual factors explain SC formation for Chile, but their relative importance varies for each kind of participation. Our second application compares individual-level determinants of SC formation among the largest democracies in the Americas. Gender, age, education, and race show heterogeneous effects across countries. Overall, negative interpersonal trust shocks generate participation increments, and possibly motivate engagement in trustworthy networks. Idiosyncratic factors behind participation and trust are positively correlated, suggesting a common SC stem that manifests in multiple ways. Hence, our empirical approach to SC formation uncovers factors hidden by assumptions in some previous literature.

Sign in / Sign up

Export Citation Format

Share Document