For_Meg_20150803.R

# In response to: https://twitter.com/duffy_ma/status/628300075110936576

df <- data.frame(x=1:10, y = 1:10)
library("plyr")

df$y <- mapvalues(df$y, from = c(5), to = c(NA)) # make sure to keep `y` column numeric
lengthdata <- length(df$x)

for(i in 1:(lengthdata-1)) {
  if (is.na(df$y[i])) {  # use is.na to test for NAs
    df$y[i] <- ((df$y[i-1]+df$y[i+1])/2)
  } else {
    df$y[i] <- df$y[i]
  }
}