Intelligence Refinery

A cool way to visualize time-series data in R.

Finding which variables in a dataset are most “similar”, in some way, to the outcome variable of interest can be a very useful first step in understanding the dataset and planning the next steps. The R package ClustOfVar provides an implementation for this purpose in the mixedVarSim() function.

Sorting a column in the dataframe does not seem to be sufficient to get a barplot that is also sorted by value. Instead, the column needs to be sorted within the ggplot() function call:

ggplot(data=df, aes(x=reorder(colX, -colY), y=colY)) + geom_bar(stat="identity")