.ipynb .pdf

repository open issue

Binder

Dynamic Modelling Introduction

The goal of this section is to analyse multiple annual reports per company to understand how companies may have changed in business direction or target industries over multiple years.

We explore two types of dynamic modelling:

1. Look at differences year over year in word embeddings generated by TD-IDF, Word2Vec and Doc2Vec. We take the difference between the word embedding vectors of two distinct years and evaluate the addition/reduction in certain vocabulary.

   • This work can be found here

2. Look at differences year over year in topics generated by Non-Negative Matrix Factorization(NMF), Latent Dirichlet Allocation (LDA), and Latent Semantic Analysis (LSA). We take the difference between the document-topic vectors and evaluate the addition/reduction in certain topics.

   • This work can be found here

previous

LSA (Latent Semantic Analysis)

next

Dynamic Company Embedding