The functionality is as follows:
- The t-SNE algorithm is a non-linear dimensionality reduction technique. Computers are easily able to interpret and process many dimensions during their computations. However, humans are only capable of visualizing two or three dimensions at a time. Therefore, these dimensionality reduction techniques come in very handy when trying to draw insights from data.
- On applying t-SNE to the 300-dimensional vectors, we are able to squash it into just two dimensions to plot it and view it.
- By specifying n_components as 2, we let the algorithm know that it has to squash the data into a two-dimensional space. Once this is done, we add all the squashed vectors into one giant matrix named all_word_vectors_matrix ...