Why do I get AttributeError: 'numpy.ndarray' object has no attribute 'replace' in python?

Question

I am trying to standardize my data in my csv file:

import pandas as pd
my_city = pd.read_csv('sample4_addlink.csv')['City'].unique()

my_city
Out[70]: 
array(['Lancaster', 'Canton', 'Edison', ..., 'Upton', 'Irvington',
       'El Cerrito'], dtype=object)

Now as you see in my_city, there are some cities with space between names like El Cerrito. I want to replace the space with an underscore

my_cities =my_city.replace(" ", "_")
Traceback (most recent call last):

  File "<ipython-input-71-53c4a0662dd7>", line 1, in <module>
    my_cities =my_city.replace(" ", "_")

AttributeError: 'numpy.ndarray' object has no attribute 'replace'

I get this error when I use the .replace function. What is the best way to get around this?

Answer 1

It's because unique returns an ndarray or ExtensionArray object depending on the data type of the input series.

Use drop_duplicates and the str accessor instead:

my_city = pd.read_csv('sample4_addlink.csv')['City'].drop_duplicates().str.replace(' ', '_')