MULTIVARIATE TECHNIQUE

Multivariate Technique: Factor Analysis

Multivariate Technique: Factor Analysis

Introduction

The study is related to multivariate techniques which include factor analysis, multi-dimensional scaling and cluster analysis; however, the particular focus of the study will be on the factor analysis. Moreover, in particular relation to high-end market research, it is imperative and essential that the upper management of the company should have its focus on the multivariate factor analysis. The basis of this statement is that multivariate technique that is the factor analysis is considered as an important statistical technique in comparison to cluster analysis and multi-dimensional scaling; as it helps in identifying important variables in the research to consider from numerous variables, which provide more accurate and reliable results to the research.

Discussion

Factor covers a set of principal component analysis, and correspondence analysis which have the specifications in common. It provides the ability to run with only one reading of the correspondence factor analysis of data, scalar products, standard products scalar co-variances and correlations. For each analysis, factor analysis builds a matrix representing the relationships between variables and computes its eigen values and eigen vectors. It then calculates the factors of “if” and “variable” giving to each “case” and “variable” their ordinates representation of their qualities and their contributions to the factors (Izenman, 2008).

Variables or cases upon which the procedure is performed, the factorial decomposition takes place that is they are used in computing the matrix of relationships. One may also search for a representation of other variables or cases within factors corresponding to the active variables. Such variables or cases having no influence on the factors are called variables liabilities.

Furthermore, the multi-dimensional scaling is a statistical technique which focuses on detecting the hidden variables, which are difficult to be observed directly and these hidden variables often explain the similarities and differences between the two objects. Moreover, the objective of multi-dimensional scaling analysis is to rearrange the distribution of the variables to study the small significant differences in explaining similarities or dissimilarities between these variables; however, it does not entail large set of variables like factors analysis (Cudeck and MacCallum, 2007).

Cluster analysis can be used in finding tree structure in objects in place of evolution in time, the group objects that are characterized in data matrix or dissimilarity and partition a data matrix before building a model for it. However, the factor analysis is also important, whose application in the study is based on the following:

Applications of Factor Analysis

Identifying the Fundamental Factors

Factor Analysis permits to get insight to categories;

Clustering if the variables into homogeneous sets;

It creates new variables that are the factors.

Screening of the Variables

Helpful in regression as it helps in resolving the multi-collinearity;

Factor analysis classifies the groupings in order to allow to select one variable to represent other variables.

Summary

Factor analysis allow in explaining several variables using a few factors.

Sampling of Variables

Factor analysis is useful in selecting small set of variables of representative variables from the larger group of variables.



Clustering of Objects

Factor analysis is useful in putting objects that also include the ...