熊猫：读取csv时不要重命名列

Question

Is it possible to avoid automatic naming of columns with empty names (resulting for instance in “Unnamed: 13”) when reading data with pandas.read_csv?

• Example

Name row in csv file:

name_1;name_2;;name_4

Names generated by read_csv:

["name_1", "name_2", "Unnamed: 3", "name_4"]

Desired names:

["name_1", "name_2", "", "name_4"]

Answer 1

You can rename columns after loading the CSV:

def rename(col):
    if col.startswith("Unnamed: "):
        return ""
    else:
        return col

data.columns = [rename(col) for col in data.columns]

I would not recommend it, and you should make sure there are no actual columns starting with "Unnamed: ", but otherwise this should work for you.

Answer 2

After you read it, you could do something like

df.columns = [x if not x.startswith('Unnamed') else i for i,x in enumerate(df.columns)]

That will replace the unnamed columns with an integer corresponding to that columns place. If you really want it to just be blank, you could do

df.columns = [x if not x.startswith('Unnamed') else "" for x in df.columns]