#Cover letter classification

###What? I've been applying to jobs. Many jobs! I started thinking, is there a way to tell which cover letters are more likely to get me hearing back from a company? The answer is, probably, and here's a shot at a model that can classify cover letters as successful (1) or unsuccessful (0). Eventually, I will use it to come up with an accuracy rate in classifying my own 30+ cover letters as 1 or 0.

In the file below, I am have a Multinomial Naive Bayes classifier. It is built using the specifications outlined in this page from Stanford's NLP department.

###Goal: Successfully classify a cover letter as successful (hear back from a company) or unsuccessful (have yet to hear back...).

###How to use:

1. Make a directory containing all of your cover letters. These must be .docx files.
2. Un-comment lines 3-9 in cover_letter_classification.R, and comment line 10.
3. Append a 0 or a 1 to the filename, preceding the .docx part. This will indicate the success of the cover letter. For example, Operations_Analytics_Analyst_1.docx
4. Run cover_letter_classification.R. Input the path to the directory that you made in Step 1.
5. If you're curious about the likely efficacy of a certain cover letter, try using the test_doc_classify function on a specific document. Otherwise, if you're curious about the leave-one-out predictions using the cover letters in your directory, these are saved in an R data.frame called predictions.