Hi,

If I understood correctly, you can just your similarity matrix and join it to the dataframe.

>>> import pandas as pd

>>> similarity_matrix = pd.np.random.rand(3,3)

>>> sample_df = pd.DataFrame({'x': [1,2,3], 'y': [4,5,6]})

>>> sample_df.join(pd.DataFrame({ind:[item] for ind,item in enumerate(similarity_matrix.ravel())}))

x y 0 1 2 3 4 5 6 \

0 1 4 0.178963 0.610631 0.892858 0.195364 0.125533 0.659616 0.707244

1 2 5 NaN NaN NaN NaN NaN NaN NaN

2 3 6 NaN NaN NaN NaN NaN NaN NaN

7 8

0 0.245501 0.616955

1 NaN NaN

2 NaN NaN

Cosine similarity in the context of text processing is well described here:

http://billchambers.me/tutorials/2014/12/22/cosine-similarity-explained-in-python.html

You can try these lines of code as well:

>>>import numpy as np

>>>import pandas as pd

>>>from sklearn.metrics.pairwise import cosine_similarity

>>>existing_set = pd.DataFrame(np.random.randint(0, 2, 2))

>>>cosine_similarity(existing_set)

This could help maybe http://scikit-learn.org/stable/modules/generated/sklearn.metrics.pairwise.cosine_similarity.html