Literal nan string stored in dataframe using json_normalize for null value in string metadata column

[joetl]

Code Sample, a copy-pastable example if possible

# Your code here
import pandas as pd
data = [{"name": "test","address": [{"add": "add1"}]},{"address": [{"add": "add2"}]}]
df = pd.io.json.json_normalize(data,"address","name",errors="ignore")
df

Problem description

In Json_normalize if the metadata column is nullable with option errors set as ignore the null values in dataframe are literal nan string and not numpy.nan

Expected Output

Expecting to see numpy.nan for null metadata string column values instead of nan string

Pandas version 0.24.1

[WillAyd]

Makes sense. Investigation and PRs are always welcome

[antoineviscardi]

Hi there,

I can take care of this. Would we want this to be the only behavior or optional by introducing a new named parameter?

[WillAyd]

@antoineviscardi this should be the only behavior