10 More Tips and Tricks for Pandas

Ten quick and useful tips and tricks for Pandas

4 min readFeb 24, 2024


The sample method lets you get a random sample of the original data frame. You can specify the number of elements (usually rows, but they can also be columns) you want in the sample or the fraction of the total elements in the data frame. The returned sample will be shuffled.

import pandas as pd

df = pd.DataFrame({'num_legs': [2, 4, 8, 0],
'num_wings': [2, 0, 0, 0]},
index=['falcon', 'dog', 'spider', 'fish'])



You can change column names directly using the columns property of the DataFrame object. However, if you only want to replace specific column names you can simply use the rename method.

df.rename({'num_legs': '#legs'}, axis='columns')


The pop method removes a column from the data frame and returns it as a Series.

The original DataFrame after removing the num_wings columns

Use column name together with iloc

If you need to use iloc but also want to use the column name instead of its index, you can do it by using

df = pd.DataFrame({'num_legs': [2, 4, 8, 0],
'num_wings': [2, 0, 0, 0]},
index=['falcon', 'dog', 'spider', 'fish'])

df.iloc[[1, 2], df.columns.get_indexer(['num_wings'])

Pay attention, if the name passed to the get_indexer is not in the data frame it will return -1 and will give you the wrong column. Suppose we use the pop method as before so the num_wings column no longer exists, then we get

You can see we got the column num_legs instead of num_wings. If you want to get one specific column you can use instead the get_loc method which raises an error if the column does not exist.

Check if two series are similar

The function pandas.testing.assert_series_equal can help you test if the two series are similar. It is similar to the equals method but it is more flexible. For example, the equals method requires the two series to be of the same type, but with the other function, we can check just for the values.

df = pd.DataFrame({"A":[1, 2, 3, 4], 
"B":[1.0, 2.0, 3.0, 4.0],

pd.testing.assert_series_equal(df["A"], df["B"], check_names=False, check_dtype=False)
df['A'].equals(df['B']) # return False

The assert_series_equal can also get a tolerance value to check if the two series are “close enough”. You should remember that this is an assertion function which means it raises an error if there is no similarity.

Convert string to numeric data

There are several ways to convert numeric data that is stored as a string to numeric data in pandas. The popular way, probably, is the to_numeric function. One advantage this function has is that it can also convert stings which are not present numerical data. In such a case, it converts the string to NaN.

df = pd.DataFrame({"A":['1', '2', '3', 'banana'], 
"B":[1.0, 2.0, 3.0, 4.0],
df.apply(pd.to_numeric, errors = "coerce")

Find the last occurrence

We can check the last time an athlete won a competition in the following way.

df = pd.DataFrame({'year': [1992, 1993, 1994, 1995, 1996, 1997, 1998],
'athlete_name': ['A', 'B', 'A', 'A', 'C', 'B', 'B']})

Control the display options

Pandas lets you control the display option (along with other parameters) using the set_option function. For example, you can control the number of rows that will be shown when a data frame is printed to the display.


You can also reset the options you changed by the reset_option function. You can see a lot of examples of such options here.

Data frames for testing

You can use the pandas.util.testing module to create data frames for testing. For example, to create a data frame with several different data types you can use


Check the memory usage for each column

You can use memory_usage function to see the amount of memory each column uses.

df.memory_usage(deep = True)