What's the best way to insert columns in a pandas Dataframe when you don't know the exact number of columns? - pandas

Question

I have an input dataframe.
I have also a list, with the same len as the number of rows in the dataframe.
Every element of the list is a dictionary: the key is the name of the new column, and the value is the value to be inserted in the cell.
I have to insert the columns from that list in the dataframe.
What is the best way to do so?
So far, given the input dataframe indf and the list l, I came up with something on the line of:
from copy import deepcopy
outdf = deepcopy(indf)
for index, row in indf.iterrows():
e = l[index]
for key, value in e:
outdf.loc[index, key] = value
But it doesn't seem pythonic and pandasnic and I get performance warnings like:
<ipython-input-5-9dde586a9c14>:8: PerformanceWarning: DataFrame is highly fragmented. This is usually the result of calling `frame.insert` many times, which has poor performance. Consider joining all columns at once using pd.concat(axis=1) instead. To get a de-fragmented frame, use `newframe = frame.copy()`

Answer 1

If the sorting of the list and the data frame is the same, you can convert your list of dictionaries to a data frame:
mylist = [
{'a':1,'b':2,'c':3},
{'e':11,'f':22,'c':33},
{'a':111,'b':222,'c':333}
]
mylist_df = pd.DataFrame(mylist)
a
b
c
e
f
0
1
2
3
nan
nan
1
nan
nan
33
11
22
2
111
222
333
nan
nan
Then you can use pd.concat to merge the list to your input data frame:
result = pd.concat([input_df, mylist_df], axis=1)
In this way, there is always a column created for all unique keys in your dictionary, regardless of they exist in one dictionary and not the other.