Tips – difference between factor and cluster analysis
Last week someone asked me about the difference between factor- and cluster analysis?
Here is my answer, that is normally the case:
Factor analysis is a help to group variables.
Cluster analysis is a help to group individuals (cases).
Some important background information that I can share with you is following:
Factor analysis is based of correlations between variables, and high loadings to a factor is the same that this variable belong to this factor. And factor is the same as the group of some variables. A high loading is over 0.6. If a variable is loading just around for example 0.3, 0.4 on more than 2 factors you can say that it’s noisy and can be deleted from the factor analysis.
Cluster analysis is building up groups of individuals/cases by measuring distances between individuals, and the distances is based of variables you choose. A good tip is to choose variables that is not correlated too much to each other. If the variables have very different scales (ex age and salary) I recommend to standardize the variables.
Many times it’s good to start with a factor analysis, and then continue with cluster analysis. One benefit could be to choose one variable from each factor (Group of variables) to the cluster analysis, as they are not correlated to each other.
Do you want to learn more about factor and cluster analysis on your own material, just ask your contact at Crayon to get more information how we could have some consulting hours together through the web.
If you want to learn more basic statistics I can recommend the classes with trainings that we can offer, read more on the tab: “Training“.