Non-Supervised Clustering of Health-Related Behavior

Using self-reported online survey data, we aim to define clusters using various unsupervised machine learning algorithms (k-means, expectation–maximization, Latent Dirichlet Allocation, Finite Mixture Model, Self-organizing maps, non-supervised k-nearest neighbors, t-Distributed Stochastic Neighbour Embedding, etc.). Datta-Datta-method and RAND index will be calculated to assess the stability of various clustering methods.

This study is part of an OSF pre-registred project: https://osf.io/zwktc