Vijay, just in short: Cluster analysis is concerned with grouping a set of objects (subjects, persons) in such a way that objects in the same group (cluster) are more similar to each other than to those in other groups.

Factor analysis is a method used for data reduction purposes. This method is concerned with grouping a set of variables and not objects. The basic idea is to represent a set of variables by a smaller number of variables, called factors.

Hope this helps, Karin

Vijay: I can give you an example in which I used Factor Analysis along with a Cluster Analysis in the same study. My goal was to study possible associations between over twenty variables (demographic; environmental; health related) on cancer rates. We first used a factor analysis to identify "factors", which reduced the dimensionality of the problem, and which resulted in "factor scores", which became our "new data sets". Then, we used each factor score data set in a spatio-temporal cluster analysis of some cancer type rates. This approach gave us results that were quite interesting, such as "populations with high income and high rates of smokers with high education level had a very high rate of breats cancer in a specific geographical area of some state in the USA".

Cluster analysis is sometimes referred to as dimension-less analysis, where the main  purpose is to identify a number of groups of variables in such a way that those variables in the same group are close to one another in some sense (the different definitions of closeness lead to different names of clustering methods).  Since cluster analysis does not employ coordinate systems, one cannot talk about 'dimensions' or dimensionality.  In contrast, factor analysis is to position variables (or subjects) on multidimensional coordinates in the so-called common factor space (if the common space is replaced with full space, it is principal component analysis - the name comes from placing variables or subjects in principal coordinate system) .   Once factors are obtained, it is typical to try to interpret each factor in terms of commonality of variables (subjects) in the same factor, and it is often the case that those initial axes are further rotated orthogonally or obliquely to attain the ease of interpretation of each factor.  Anyway, cluster analysis and factor analysis are totally different in terms of analysis and interpretation.

I think creativity does not happens in vacuum, it take sits elements from the existing environment and creates a new entity.  If we miss the half reality there will be definitely limited view or reality and the creativity will be half or even lesser than half. I mean a complete reality is needed for complete creation.

Almost all datasets do have cases in row 1 and variables in column 1.

You can transpose (flip) the datamatrix. (just like transpose in excel). data in rows become data in columns.

In the field of psychology many years ago this was done on datasets to perform q-factoring. in stead of factoranalysis, combining items into factors it became combining persons into groups (clusters). the reason to do this (in earlie 1980ies) was that factor analysis was implemented in statistical packages and people new factor analysis. As I remember the advice was not to continu to use q-factoring because it violates several assumptions that are given for datasets. For item scores there is normaly a Gausian distibution, and items are inter-related. the same is not true for clusters of persons. q-factoring was considered tot be a quick and dirty method to do cluster analysis. however, there are some studies that do find that results of q-factoring or cluster analysis do not differ that much. thus for some datasets it still can be done.

best advice seems to be: use factor analysis for combining items into factors, use cluster analysis for combining persons into clusters.

and for people knowing all about structural equations or path analysis it is possible to use methods that can do both simultaneously. for this method however you must know much more in advance, it is less exploratory, and more confirmatory (not conpletely so).