Question

When reading a csv file, where each cell can be strings or numerical values. Which approach should I use to read this csv file into a matrix. The tricky thing is that I may need to perform some computation on this imported matrix. If an entry is a string, I need to perform some character-based operation on this string, e.g., comparing it with another string. If an entry is an numerical value, I need to perform a add/subtract operation on it. How should I import this csv file:

testmatrix = as.character(read.csv("test.csv", sep=","))
testmatrix = as.vector(read.csv("test.csv", sep=","))

The data is like this

word1   word2   123  word3
234     456     word4  word5

Answer 1

You can read into a dataframe with one of the read.* functions but since there are no commas, read.csv doesn’t make much sense. Be sure to use ‘stringsAsFactors’ to avoid automatic factor creation. Convert to matrix to facilitate processing with vectorized functions like sum:

 dat <- read.table(text="word1   word2   123  word3
 234     456     word4  word5", header=FALSE, stringsAsFactors=FALSE)
 mdat <- data.matrix(dat)  # you get warning but it's safe to ignore them in this case.
 sum(as.numeric(mdat)[ is.numeric(mdat) ] )
#[1] NA  left in to illustrate need to use na.rm=TRUE
 sum(as.numeric(mdat)[ is.numeric(mdat) ] , na.rm=TRUE)
#[1] 813   expected result