Cluster Analysis : R software

Q1. Why would you reset the seed to the random number generator prior to running K-Means?

Q.2 If you have a lot of columns describing some observation, how would you go about deciding which columns to use? Why not use them all?

Q3. What impact does outlier data have on the K-Means algorithm?

Q4. If you had a large data set (very high number of rows) and K-Means was running very slowly on your laptop, could draw a sample from the data set and run K-Means on that? What problems could come from that?

Q.5 How can cluster analysis be used to improve a regression model? What would you learn from clustering that would aid you in making decisions about your regression analysis?