We want to predict specificity, constructiveness, influence, and other attributes (subject to change) from quotes of employees giving feedback to their managers. An outline of potential steps (also open to your thoughts on this) might be: 1. take the quotes data 2. take the ratings / quantitative 3. label how constructive or specific or valuable the quotes are (1-5) (we only use 80% of it) 4. on ml side, we would create relevant auto-calculate features for all the quotes 5. collate all the features, the quotes themselves, and the quantitative data into a cvs 6. we could run various models -- regression, svr, nb, rnn, rf 7. we would test how well our models perform agains the 20% that were labeled but not used in training we could also investigate deep learning models (NN -- RNN, charCNN)
Years: Any
Location: Anywhere
Requested on: 2021-11-09
Natural language processing