Numpy np.where multiple condition

Question

I need to work with multiple condition using numpy.

I'm trying this code that seem to work.

My question is: There is another alternative that can do the same job?

Mur=np.array([200,246,372])*pq.kN*pq.m
Mumax=np.array([1400,600,700])*pq.kN*pq.m
Mu=np.array([100,500,2000])*pq.kN*pq.m
Acreq=np.where(Mu<Mur,0,"zero")
Acreq=np.where(((Mur<Mu)&(Mu<Mumax)),45,Acreq)
Acreq=np.where(Mu>Mumax,60,Acreq)
Print(Acreq)
['0' '45' '60']

Answer 1

Starting with this:

Mur    = np.array([200,246,372])*3*5
Mumax  = np.array([1400,600,700])*3*5
Mu     = np.array([100,500,2000])*3*5
Acreq  = np.where(Mu<Mur,0,"zero")
Acreq  = np.where((Mur<Mu)&(Mu<Mumax),45,Acreq)
Acreq  = np.where(Mu>Mumax,60,Acreq)

print(Acreq)

['0' '45' '60']

Try this:

conditions  = [Mu<Mur, (Mur<Mu)&(Mu<Mumax), Mu>Mumax ]
choices     = [ 0, 45, 60 ]
Acreq       = np.select(conditions, choices, default='zero')
print(Acreq)


['0' '45' '60']

This also works:

np.where((Mur<Mu)&(Mu<Mumax),45,np.where(Mu>Mumax,60,np.where(Mu<Mur,0,"zero")))

Answer 2

you can use Pandas's pd.cut() method:

generate random series of integers:

In [162]: import pandas as pd

In [163]: s = pd.Series(np.random.randint(-3,10, 10))

In [164]: s
Out[164]:
0    6
1   -3
2    6
3    6
4    7
5    7
6    3
7   -2
8    9
9    1
dtype: int32

categorize them:

In [165]: pd.cut(s, bins=[-np.inf, 2, 5, np.inf], labels=['0', '45', '60'])
Out[165]:
0    60
1     0
2    60
3    60
4    60
5    60
6    45
7     0
8    60
9     0
dtype: category
Categories (3, object): [0 < 45 < 60]