CODEWITHSUNDEEP
pythonpandassplit
andrewmcg
andrewmcg

Reputation: 23

How can I get the "index" from a look alike dict in Pandas?

I have basically the same problem of this guy: How can I get a string column that look like a dictionary and get the last item of it?. However, instead of getting the last number, I need the index. Which is not actually an index, but a string.

As shown in data, 'col_B' contains a list of strings.

Any ideas?

data =\
{'col_A': ['AA', 'BB', 'CC'],
 'col_B': ['{"0":10,"5":13,"8":20}', '{"0":2,"3":34,"5":40,"15":100}', '{"2":5,"5":19,"15":200,"20":200,"30":340}']}

df = pd.DataFrame(data)


  col_A                                      col_B
0    AA                     {"0":10,"5":13,"8":20}
1    BB             {"0":2,"3":34,"5":40,"15":100}
2    CC  {"2":5,"5":19,"15":200,"20":200,"30":340}

I need to find a way of extracting the numbers 8, 15 and 30.

Upvotes: 2

Views: 43

Answers (1)

yatu
yatu

Reputation: 88276

You could parse the strings with literal_eval and index dict.keys to obtain the last key:

from ast import literal_eval

df['col_B'] = df.col_B.map(literal_eval)
df.col_B.map(lambda x: list(x.keys())[-1])
0     8
1    15
2    30
Name: col_B, dtype: object

Though depending on the version of python, the order of the dictionary may not be preserved. In such case, perhaps a regex is safer:

df.col_B.str.extract(r'"(\d+)"\:\d+}$').squeeze()
0     8
1    15
2    30
Name: 0, dtype: object

Upvotes: 4

Related Questions