Pandas is a data manipulation library in Python. loc and iloc are two important methods used for accessing data in a DataFrame.

As a user of pandas I have wanted to write down a simple explanation of using loc and iloc in pandas.

loc

This is the most useful one for me. We can retrieve the rows by index and the columns by name. The syntax is like df.loc[row_label, column_label] and

import pandas as pd

df.loc[0] # returns the first row
df.loc[0, 'A'] # returns the first row, first column

df.loc[:2]  # returns the first two rows

df.loc[0:2, 'B']  # returns the first two rows, B column

df.loc[1:, ['A', 'B']]  # returns the first two rows, A and B columns

df.loc[df['A'] > 1]  # returns rows where A is greater than 1

df.loc[df['A'] > 1, 'B']  # returns rows where A is greater than 1, B column

df.loc[df['A'] > 2, ['B', 'C']]  # returns rows where A is greater than 2, B and C columns

df.loc[df['A'] > 1, ['B', 'C']] = 0  # set values of B and C to 0 where A is greater than 1

df.loc[0, 'A'] = 10  # set the value of A in the first row to 10

df.loc[0] = 20  # set all values in the first row to 20

df.iat[0, 0]  # returns the value in the first row, first column

iloc

Although a bit less useful, if we know the index of the columns we can do exactly the same as with loc but provide the column positions instead.

import pandas as pd

data = {'A': [1, 2, 3], 'B': [4, 5, 6]}

df.iloc[0]  # returns the first row

df.iloc[0, 0]  # returns the first row, first column

df.iloc[:2]  # returns the first two rows

df.iloc[0:2, 1]  # returns the first two rows, second column

df.iloc[1:, [0, 1]]  # returns the first two rows, first and second columns

Use loc when you want to access data by labels and iloc when you want to access data by positions.

Bonus: at and iat

When writing this post, I found two more methods that are similar to loc and iloc. They work just like loc and iloc but are focused on retrieving a single value.

import pandas as pd

data = {'A': [1, 2, 3], 'B': [4, 5, 6]}

df = pd.DataFrame(data)

df.at[0, 'A']  # Output: 1

df.iat[0, 0]  # Output: 1