dataset - How much data do I need for recommender system? -


i have develop personality/job suitability online test hr department. basically, users answer questions, on scale of 0-10 example, , after 50 questions, want translate rating in 5 different personality/ job suitability characteristics.

i don't have real data start with, first, worth use recommendation engine mymedialite (github). how many samples need train decent performance?

i built training course recommender, doing , hand-weighted sum each question increased weight of several courses related question. expert system, built feed-forward neural network, tuned weights based on knowledge of questions , courses' content.

i time around use recommender system, i'm wondering how many times have take 50 question test, , assign results manually. 100 examples do? possible. 1000 long. how can know ahead of time?

though useless, want not possible give definite number. should focus on learning curve when adding new samples.

you can process samples hand , engine on parallel, , compare result given both. once measurement e.g. recall , precision of result given engine achieve expectation, enough samples.

hope helpful!


Comments

Popular posts from this blog

ios - UICollectionView Self Sizing Cells with Auto Layout -

node.js - ldapjs - write after end error -

DOM Manipulation in Wordpress (and elsewhere) using php -