R Function Not Returning as Expected - r

Question

I've been working on a function in R, and looking to return a data.frame. However, when I run, it return 0 rows/0 columns. I did add a "print" in the if statement, and that works, so I know it's performing as intended within the loop and if statement.
Here is the function:
predict.next.word <- function(word, ng_matrix){
if(ncol(ng_matrix)==3){
for (i in 1:100){
ngram_df <- data.frame()
first_word <- ng_matrix[,1][i]
second_word <- ng_matrix[,2][i]
count_word <- ng_matrix[,3][i]
if (word[1] == first_word && !is.na(first_word)){
matched_factor <- structure(c(second_word, count_word), .Names = c("predicted", "count"))
matched_df <- as.data.frame(as.list(matched_factor))
ngram_df <- (rbind(ngram_df, matched_df))
ngram_df <- transform(ngram_df, count = as.numeric(count))
print (ngram_df) # this works great, but not intention of function
}
}
return (ngram_df)
}
}
Here is a sample of when I call it:
test_bigram_word <- c("a")
predict.next.word(test_bigram_word, bigram_index)

Answer 1

Solved it by moving the initializing of the dataframe above the for loop, like this:
predict.next.word <- function(word, ng_matrix){
if(ncol(ng_matrix)==3){
ngram_df <- data.frame()
for (i in 1:100){
first_word <- ng_matrix[,1][i]
second_word <- ng_matrix[,2][i]
count_word <- ng_matrix[,3][i]
if (word[1] == first_word && !is.na(first_word)){
matched_factor <- structure(c(second_word, count_word), .Names = c("predicted", "count"))
matched_df <- as.data.frame(as.list(matched_factor))
ngram_df <- (rbind(ngram_df, matched_df))
ngram_df <- transform(ngram_df, count = as.numeric(count))
print (ngram_df) # this works great, but not intention of function
}
}
return (ngram_df)
}
}