I have a text file containing tweets and I'm required to count the number of times a word is mentioned in the tweet. For example, the file contains:

Apple iPhone X is going to worth a fortune
The iPhone X is Apple's latest flagship iPhone. How will it pit against it's competitors?

And let's say I want to count how many times the word iPhone is mentioned in the file. So here's what I've tried.

cut -f 1 Tweet_Data | grep -i "iPhone" | wc -l

it certainly works but I'm confused about the 'wc' command in unix. What is the difference if I try something like:

cut -f 1 Tweet_Data | grep -c "iPhone"

where -c is used instead? Both of these yield different results in a large file full of tweets and I'm confused on how it works. Which method is the correct way of counting the occurrence?

up vote 21 down vote accepted

Given such a requirement, I would use a GNU grep (for the -o option), then pass it through wc to count the total number of occurrences:

$ grep -o -i iphone Tweet_Data | wc -l

Plain grep -c on the data will count the number of lines that match, not the total number of words that match. Using the -o option tells grep to output each match on its on line, no matter how many times the match is in the line.

wc -l tells the wc utility to count the number of lines. After grep puts each match in its own line, this is the total number of occurrences of the word in the input.

If GNU grep is not available (or desired), you could transform the input with tr so that each word is on its own line, then use grep -c to count:

$ tr '[:space:]' '[\n*]' < Tweet_Data | grep -i -c iphone

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.