How to create a nested dictionary from a given dataframe?-CodePudding

I got a df like this one:

level       profile    chest_gold       chest_silver        chest_bronze
1           a          TRUE             FALSE               TRUE
2           a          FALSE            FALSE               TRUE
3           a          FALSE            TRUE                TRUE

I want to obtain a dictionary which uses as key the level and the profile to return something like this, converting the TRUE/FALSE in 1/0:

d[profile][level] = [1, 0, 1]  #the chest result

for example:

d['a'][1] = [1,0,1]
d['a'][2] = [0,0,1]
d['a'][3] = [0,1,1]

How can I do that?

P.s. if you leave a solution please leave also a little comment to explain the answer!

CodePudding user response：

Check a nested dictionary comprehension. The inner comprehension builds dictionary where keys are levels and outer comprehension builds a dictionary where the keys are profiles.

output = {k: {i: v for i, *v in d.set_index('level').filter(like='chest').astype(int).to_records()} for k, d in df.groupby('profile')}

{'a': {1: [1, 0, 1], 2: [0, 0, 1], 3: [0, 1, 1]}}

CodePudding user response：

I think using itertuples() to loop over rows would be more efficient. setdefault() helps to build a nested dictionary by inserting a profile as key with an empty dictionary (which is filled in with level-chest values as the loop goes on)

# build a nested dictionary while iterating over rows of df
d = {}
for lvl, profile, *chest in df.mul(1).itertuples(index=False):
    d.setdefault(profile, {})[lvl] = chest
d

# {'a': {1: [1, 0, 1], 2: [0, 0, 1], 3: [0, 1, 1]}}