[FEA] Use category dtype in PropertyGraph DataFrames

[VibhuJawa]

trafficstars

Describe the solution you'd like and any additional context

We should use category dtype for _TYPE_ column rather than string for _edge_prop_dataframe and _vertex_prop_dataframe.

This will have following benefits:

1. Saves on memory Consumption (See below example for motivation)
2. Faster because of numerical dtypes
3. Also aligns us to Graph Frameworks like DGL who also use category like dtypes to save _TYPE_ under the hood.

Memory Usage Motivational example:

edge_prop_dataframe



_SRC_ | _DST_ | _EDGE_ID_ | _TYPE_ | user_id_1 | user_id_2 | merchant_id | stars | relationship_type | user_id | volume | time | card_num
-- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | --
89216 | 78634 | 0 | referrals | 89216 | 78634 | 11 | 5 | <NA> | <NA> | <NA> | <NA> | <NA>
89216 | 78634 | 0 | referrals | 89216 | 78634 | 11 | 5 | <NA> | <NA> | <NA> | <NA> | <NA>
89216 | 78634 | 0 | referrals | 89216 | 78634 | 11 | 5 | <NA> | <NA> | <NA> | <NA> | <NA>

get_size(edge_prop_dataframe['_TYPE_'].memory_usage())
'3.0 GB'

get_size(edge_prop_dataframe['_TYPE_'].astype('category').memory_usage())
'1.17 GB'

[github-actions[bot]]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.