Open my_thesis/main.pdf
Description of an unsupervised model developed to discover User Habits (working hours, working days, sports times, family visits, ect...) from his smartphone data. This model is a latent variable probabilistic model inspired form LDA (Latent Dirichlet Allocation).
The model was tested on a Sony smartphone dataset. The report also discusses the results of the model and shows that it outperforms current state of the art models (LDA, LCBMF).
The model was developed and tested using Python and Matlab.