I'm working on preparing/analysing this dataset of Russian propaganda ads, in which there are columns which contain lists.
'interests_categories', for example, has a row which contains ['Unknown', 'Geographic', 'LGBTQ'].
I aim to compare classifiers and feature selection methods, create pipelines, and experiment with clustering and manifold learning. However, my understanding is that these lists will be tricky to deal with.
The dataset already contains some pre-made regex columns (including the above example), which concatenates all the values in the list with underscores. I have also found the possibility of converting these lists to tuples instead.
Is one of these approaches recommended above the other?