Contains tests of combining different modules to get tweets, vectorize them, and cluster them.
Uses the sequential version of Birch algorithm to cluster tweets.
Use the MeanShift algorithme on the tweet stream after the vectorization using sklearn - that we later improoved.
Vectorization then clustering of a predefined corps chosen so that we can guess its number of clusters.
K-means clustering with tweeter input, to test actual use of TensorFlow code
CEM clustering with tweets under JSON format, used to test the algorithm under different kinds of input.