Python 3：按最后一行的列值对 DF 的字典进行排序

Question

Dict of df's:

years = [2017, 2018, 2019]
dfcols = ['Count', 'Amount']
dataframedict = {'CatA': pd.DataFrame(data=[[1, 5], [2, 6], [3, 7]], columns=dfcols, index=years), 
                'CatB': pd.DataFrame(data=[[4, 8], [5, 9], [6, 19]], columns=dfcols, index=years), 
                'CatC': pd.DataFrame(data=[[7,11], [8,12], [9,13]], columns=dfcols, index=years)}

Need a list of dict keys sorted descending by 'Amount' in 2019, so ['CatB', 'CatC', 'CatA']

Tried, following this variations of:

for k in sorted(dataframedict, key=lambda df: df.Amount[-1], reverse=True):
    print(k)

Getting error: AttributeError: 'str' object has no attribute 'Amount'

Thanks

Answer 1

If you just want a list, you can use a list comprehension and the .items method of the dictionary to retrieve the key and dataframe

[
    k for k, df 
    in sorted(dataframedict.items(), key=lambda x: x[1].Amount.iloc[-1], reverse=True)
]
# returns:
['CatB', 'CatC', 'CatA']

Answer 2

You can convert your dict to a sort list of tuples:

for k, v in sorted(dataframedict.items(), key=lambda x: x[1].Amount.iloc[-1], reverse=True):
    print(k)

Output:

CatB
CatC
CatA

Note that I am passing .items() to the sorted . Each one is a tuple of length 2 ( x[0] is the key and x[1] is the value/df).

Now, regarding your error: When you pass a dict in sorted, the keys of the dict are considered:

>>> a = {1: "ad", 0: "ads"}
>>> sorted(a, key=lambda x: x)
[0, 1]

Thus you are passing strings in the lambda