How can I add a column to an existing filled dataframe

Question

I tried to add value to a new column for an existing data frame:

val_df["perplexity"][10]=14

There is no column named perplexity and I thought this will create the column. I guessed this answer says the same, but I got the following error:

KeyError: 'perplexity'

So if you want to add new column called perplexity, you need to specify the values. What are those values? — Dani Mesejo, Mar 06 '19 at 19:20
@DanielMesejo they must be set in a loop, however I can set a default value of `0` — Ahmad, Mar 06 '19 at 19:21
Then I guess you are just find doing `val_df["perplexity"] = 0` and then setting the other values with the for loop. — Dani Mesejo, Mar 06 '19 at 19:22

Supratim Haldar · Accepted Answer · 2019-03-06T19:59:34.083

2

Since 'perplexity' column does not exist yet, a specific index (row) cannot be accessed. So, create the column and assign a default value, then update specific row's value.

val_df["perplexity"] = ''
val_df["perplexity"][10] = 14

EDIT: As suggested by @ALollz below, chained indexing should be avoided to avoid the SettingWithCopy warning. The following code is recommended.

val_df.loc[10, 'perplexity'] = 14

edited Mar 06 '19 at 19:59

answered Mar 06 '19 at 19:29

Supratim Haldar

2,376
3
16
26

5

chained assignment should be avoided. Use `.loc`: `val_df.loc[10, 'perplexity'] = 14` – ALollz Mar 06 '19 at 19:40
1

You are right, thanks for pointing it out. I have added this as an edit. – Supratim Haldar Mar 06 '19 at 19:52

How can I add a column to an existing filled dataframe

1 Answers1