Floor. With math.floor we reduce a number so that the fractional part is removed. Floor() will make all numbers smaller. Negative values will become more negative.

number = 78.6# This will not work.
result = floor(number)
Traceback (most recent call last):
File "C:\programs\file.py", line 5, in <module>
result = floor(number)
NameError: name 'floor' is not defined

Negative floors. With negative numbers, math.floor has the same logical result as with positive ones. Numbers are always reduced. Negative numbers will become more negative.

Python program that uses math.floor on negative numbers

import math
# Use math.floor on a negative number.
result = math.floor(-1.1)
print(result)
result = math.floor(-1.9)
print(result)
-2
-2

Benchmark, floor dictionary. I wanted to test whether a dictionary lookup could be faster than a math.floor call. I found math.floor is fast—much faster than calling get().

Version 1 This version of the code uses math.floor to get the floor of each number.

Version 2 Here we look up a value in a dictionary to get a cached floor value for a number.

Result Using a dictionary to memoize (cache) the result of math.floor is a big slow down. Just use math.floor directly.

Python program that benchmarks math.floor, dictionary

import time, math
# Floor dictionary.
floor_dict = {100.5: 100}
print(time.time())
# Version 1: use math.floor.
for i in range(0, 100000000):
y = 100.5
z = math.floor(y)
if z != 100:
print(z)
break
print(time.time())
# Version 2: use dictionary lookup, get method.
for i in range(0, 100000000):
y = 100.5
z = floor_dict.get(y)
if z != 100:
print(z)
break
print(time.time())
1454633830.142
1454633830.727 math.floor = 0.59 s
1454633839.844 floor_dict.get = 9.12 s
PyPy3 used

A review. In Python 3 we find many built-in functions like abs and round. With floor, though, we have to use the math module with an import statement.