Suppose I have a DataFrame including following columns "NAME", "SURNAME", "AGE" and I would like to create one object for each row, including those column values as its variables.
person = ConvertRow2Object(frame.iloc[0,:])
print person.NAME //outputs Gary
How can I do it with a generic solution to any DataFrame with any kind of column names and data types?
You can convert the whole thing to a numpy recarray, then each record in the array is attributed:
people = frame.to_records()
person = people[0]
print person.NAME # etc...
Using a namedtuple also seems to work:
from collections import namedtuple
Person = namedtuple('Person', frame.dtypes.index.tolist())
person = Person(*frame.iloc[0,:])
print person.NAME # etc...
// ...? Is that suposed to be a C style comment?
....This technique of creating dictionary objects and passing that as init argument worked for me. This is also more generic in that you don't need to type in the key / property names
# AppInfo = Class which I wanted to create instance objects for each row in df
class AppInfo:
def __init__(self, attr_dict):
if attr_dict is not None:
for key, value in attr_dict.items():
setattr(self, key.replace(' ', '_'), value)
# in my code this methods creates a list of AppInfo objects created from the dataframe
def get_app_infos() -> List[AppInfo]:
df = data.query_from_db()
[AppInfo(a) for a in df.to_dict('records')]
person['name'])?