Karthik and Richard Session
===

Next thing to do would be to cache the data

Put all this in a file called load.r

# read sql data into R.
# if R data.frame is now named data, then
# Delete that row with the really large # downloads

data=subset(data,downloads<=300)

save(data,file="myexperiment_data")

# Next time you want to read the saved data, 

# Then in another file called run.r, this line below would be up top and we only run this file.

load('myexeriment_data")

# so fit generalized linear models rather than linear regression
# Do a cluster analysis on attritubes that Richard would list here