Last Updated on July 14, 2022 by Jay
In this tutorial we’ll cover how to round values up, down, to the nearest numbers in pandas.
Let’s use the below simple dataset for the demonstration:
import pandas as pd
import numpy as np
df = pd.DataFrame({'a':[3.14159, 1.234, 3.456, 10.111, -3.3],
'b':[12345, 90010, 7600, -9876, 15671.3]})Pandas round() Syntax
Note this is pandas’ round() method, not the Python built-in round() function. That said, both round() work similarly.
DataFrame.round(decimals=0)Both DataFrame and Series classes have round() methods that work exactly the same.
Round Values To N Decimals
Simply pass in an integer value into the round() method to round values to the desired decimals. For example, to round to 2 decimals:
df
a b
0 3.14159 12345.0
1 1.23400 90010.0
2 3.45600 7600.0
3 10.11100 -9876.0
4 -3.30000 15671.3
df['a'].round(2)
0 3.14
1 1.23
2 3.46
3 10.11
4 -3.30
Name: a, dtype: float64Round Values Up In Pandas
To round values up, we’ll need to leverage the numpy.ceil() method, which returns the ceiling (i.e. rounded-up number) of the input. The ceil() method can take either one or multiple input values. The below two ways return the same results:
np.ceil(df['a'])
0 4.0
1 2.0
2 4.0
3 11.0
4 -3.0
Name: a, dtype: float64
df['a'].apply(np.ceil)
0 4.0
1 2.0
2 4.0
3 11.0
4 -3.0
Name: a, dtype: float64In the above code, note the df.apply() takes a function as an input.
Round Values down
Of course, there’s also a numpy.floor() method that returns the floor (i.e. rounded-down number) of the input. The syntax is similar to the above example.
df['a'].apply(np.floor)
0 3.0
1 1.0
2 3.0
3 10.0
4 -4.0
Name: a, dtype: float64Round Values to Nearest Thousands
The pandas round() method actually allows for negative numbers as input. The negative input specifics the number of positions to the left of the decimal point. For example:
- round(decimals = -1): round to the nearest tens
- round(decimals = -2): round to the nearest hundreds
- and so on
To round to the nearest thousands, we’ll just set decimals = -3.
df['b'].round(-3)
0 12000.0
1 90000.0
2 8000.0
3 -10000.0
4 16000.0
Name: b, dtype: float64Round A DataFrame With Different Criteria
The decimals argument in the round() method can be either an integer value, or a dictionary. This makes rounding multiple columns easy at the same time.
We can round the first column to 2 decimals, and round the second column to the nearest thousands as the following:
df.round({'a':2, 'b':-3})
a b
0 3.14 12000.0
1 1.23 90000.0
2 3.46 8000.0
3 10.11 -10000.0
4 -3.30 16000.0Additional Resources
Pandas apply() method