Automatically deskew before machine learning in R

A common data pre-processing step in R is to deskew data, which is where if a histogram shows a lopsided distribution, apply a function such as log() before fitting a model. If there are a large number of columns, it can be tedious to eyeball each histogram, and manually substitute offending columns with their log() counterparts.

Helpfully, the e1071 package (notably for its support vector machine algorithms) provides a handy function to measure the skewness of data, called skewness(). Below is a function to automatically deskew an entire range of columns of a data frame.

deskew <- function(df, mincol=1, maxcol=ncol(df), threshold=1.10) {
  for (i in mincol:maxcol) {
    t <- log(1+df[[i]]-min(df[[i]]))
    if (abs(skewness(df[[i]])) > threshold * abs(skewness(t)))
      df[[i]] <- t
  }
  df
}

Deskewing data improves the performance of linear models, both regular lm()/glm() and linear svm() support vector machines. Understandably, it doesn't help with decision trees such as randomTrees().

SQL HAVING in R

Coming from a SQL background, I'm learning R, and wanted to do the equivalent of GROUP BY HAVING (well, really embedded in a subquery in order to subset the data), but the most obvious Google searches turned up nothing.  The answer is probably a no-brainer for R experts, but here it is in case future R-novices-SQL-experts Google for it.

Taking the example data set chickwts,

> data(chickwts)
> summary(chickwts)    
    weight          feed
Min.   :108.0   casein   :12
1st Qu.:204.5   horsebean:10
Median :258.0   linseed  :12
Mean   :261.3   meatmeal :11
3rd Qu.:323.5   soybean  :14
Max.   :423.0   sunflower:12

and supposing we want to exclude "low" popularity feeds that occur in the data set fewer than 12 times (yes, this is a contrived example), the below will discard those low-popularity feeds, leaving only the high-popularity feeds.

x <- sapply(split(chickwts,chickwts$feed), nrow)
chickwts <- chickwts[chickwts$feed %in% names(x[x>=12]),]