What is the difference between np.min and min?

Question

I have trying to learn Python on datacamp. When I run statistical codes with agg - like min, max - I realized that it is the same result with using np with these codes. But I can not use mean and median methods without np.

So,

is there any difference between np.min and min , np.max and max?
why don't median and mean work without numpy?

Code:

unemp_fuel_stats = sales.groupby("type")["unemployment", 
              "fuel_price_usd_per_l"].agg([np.min, np.max, np.mean, np.median])

Result:

unemployment                      fuel_price_usd_per_l  
amin   amax   mean median                 amin   amax   mean median
type  
A           3.879  8.992  7.973  8.067                0.664  1.107  0.745  0.735
B           7.170  9.765  9.279  9.199                0.760  1.108  0.806  0.803

Another result:

unemployment        fuel_price_usd_per_l  
min    max                  min    max
type  
A           3.879  8.992                0.664  1.107
B           7.170  9.765                0.760  1.108

one is plain python, the other is numpy-datastructures-aware. — Mike 'Pomax' Kamermans, Jan 19 '23 at 21:23
1 - https://stackoverflow.com/questions/10943088/numpy-max-or-max-which-one-is-faster 2 - You can use mean and median without numpy `.agg(['min', 'max', 'mean', 'median'])` — It_is_Chris, Jan 19 '23 at 21:24
See https://numpy.org/doc/stable/reference/generated/numpy.amin.html#numpy.amin, for example. `np.min` can do much more than `min` alone. — chepner, Jan 19 '23 at 21:24
`from statistics import mean, median`, not restricted to `numpy` ;) — mozway, Jan 19 '23 at 21:33
@MichaelCao yes, `sales` appears to be a pd.DataFrame and you can just do `.agg(['min', 'max', 'mean', 'median'])` and not use `numpy` -- `pd.DataFrame([[1,2,3], [2,5,6]]).groupby(0).agg(['min', 'max', 'mean', 'median'])` — It_is_Chris, Jan 19 '23 at 21:37
@It_is_Chris that is almost certainly using the `numpy` versions, it isn't clear what you mean "without `numpy`" — juanpa.arrivillaga, Jan 19 '23 at 21:47

Michael Cao · Accepted Answer · 2023-01-19T21:38:40.990

min is the builtin python function and numpy.min is another min function that is in the numpy package. They act similarly but different.

Given a single list, they will act the same:

print(np.min([1, 2, 3, 4, 5]))
print(min([1, 2, 3, 4, 5]))

1
1

However, given multiple arguments, numpy.min will throw an error as the first argument must be array-like. min will simply return the minimum of all elements.

print(min(1, 2, 3, 4, 5))

try:
    print(np.min(1, 2, 3, 4, 5))
except Exception as e:
    print(e)

1
output must be an array

numpy.min can take the minimum along a specific axis of a multi-dimensional array (which uses multiple square brackets) or the minimum of all elements. min will simply given the minimum list (though I forget how they're ordered).

twod_array = [[1, 2, 5],
              [3, 7, 5],
              [10, 5, 7]]

print(min(twod_array))
print(np.min(twod_array))
print(np.min(twod_array, axis = 0)) # Minimum of each column
print(np.min(twod_array, axis = 1)) # Minimum of each row

[1, 2, 5]
1
[1 2 5]
[1 3 5]

Hope that was enlightening. The tl;dr is that they are simply different functions with the same name. It's good to use import numpy as np to differentiate min from np.min. Using something like from numpy import min will cause ambiguity as to which function is actually being used.

Python has no builtin median or mean but numpy has numpy.median and numpy.mean

What is the difference between np.min and min?

1 Answers1