0
$\begingroup$

I have a data set that consists of different characteristics of communities. What I want to do is to see how those characteristics influence each other. As in, for example I have the income and population and other demographic data for a community, and also the level of access to some different services and I want to determine the access level to those services for the communities. I want to create a model of those communities, like what kind of factors are influencing the community access to a service - high income, high population density, education etc.

I have performed Principal Component Analysis on the data set and I've obtained the score and loading plots (the correlation between the principal components and the initial variables). What could I do next in my analysis using those? I was thinking about regression, but I am not sure exactly how to use the data from the PCA in the regression analysis.

$\endgroup$
  • 1
    $\begingroup$ Usually the output of the PCA will tell you how to linearly combine your features in such a way to reduce the number of dimensions you need to explain the variance of your data. However, it won't necessarily tell you which individual features are more important than others. If you want that, why not run a ridge regression on your dataset? That may help eliminate certain features which aren't useful in predicting your dependent variable. Also you might want to read this stats.stackexchange.com/questions/27300/… $\endgroup$ – ilanman Jun 30 '15 at 20:49
  • $\begingroup$ I did read the link you've posted before adding the question. I might try and run the ridge regression on the data set. What I actually want is to characterize those communities based on demographics and then predict their access to different services. $\endgroup$ – joh Jul 1 '15 at 8:57
  • 1
    $\begingroup$ If you don't necessarily care about the specific features and are more interested in the predictions, then you can go ahead and run a PCA and on the result significant PCs, you can run a logistic regression (for example) to make predictions. $\endgroup$ – ilanman Jul 1 '15 at 17:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.