[Pycon] [new paper] "Chiara Basei" - Recommendation Model for Ranking Matching Houses

[info a pycon.it]

Title: Recommendation Model for Ranking Matching Houses
Duration: 45 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: We address the problem of missing data for recommendation system in python 2.7 with **H2O** package. We impute the missing values by comparing two models (**Generalized Low Rank Model** & **Deep Learning with Autoencoder**), as well as presenting a new ranking formula inspired by Lucene Similarity. As a result of tuning model parameters and preforming multiple accuracy tests, Deep Learning with an Autoencoder consistently preforms better in term of precision, while GLRM performs better in term of execution time.



Tags: [u'Recommendation', u'Deep-Learning', u'GLRM', u'Autoencoder', u'H2O']