Rscript -e 'df <- read.csv("data.csv"); print(tapply(df$salary, df$dept, mean))'

tapply applies a function to groups of a vector: here it splits df$salary by the df$dept factor and computes the mean within each group, a group-by operation. The dollar sign syntax selects a column from the data frame df. The result is a named vector of per-department averages. This is base R's answer to SQL GROUP BY without loading extra packages.