R codes here include the functions in the alternating iterative algorithm for the Logistic Collaborative Model (LogCM) family -- LogCM, Similarity-regularized LogCM (LogSCM) and Pairwise-fusion LogCM (LogPCM).
main.R provide an full example of learning the toy dataset data_sample.csv with the LogCM family.
Algorithms.R includes LogCM and LogSCM functions, Algorithms_PCM.R and auxs.R include the modifications for LogPCM cases.
Logistic Collaborative Model (LogCM) is a personalized binary classification model, which can learn a distinct model for each individual in a heterogeneous population. It is inspired by the work of collaborative learning where canonical models are used to describe the low-rank structure of the personal models. For more information, please check our papers:
A Learning Framework for Personalized Random Utility Maximization (RUM) Modeling of User Behavior