Big data needs big analyses and big visualization. Probabilistic Latent Semantic Analysis (PLSA) was originally developped as an indexing tool to organize large collections of word documents from word occurences. PLSA is a dimension reduction technique that finds patterns in the dataset by probabilistically determining the 'topics' driving the word-document structure. For example, the frequent co-occurence of the words 'hollywood', 'love', and 'celebrity' could be detected in a collection of magazines as being strongly associated to a topic. Different visualization can are used to explore the relationship between topics and the word-document structure such as parallel plots.

MicrobPLSA expands Mathieu Blondel's PLSA python package by adding some analyses modules and automizing different visualization techniques.

Packages:

• numpy
• scipy
• matplotlib

Note: microbPLSA was developped in the 2.7 version of Python