How to create an numpy array of labels from a numpy array of float?

Question

For example, I have

arr=np.linspace(0,1,11)

and I want to mark numbers n<0.25 label "a" , n>0.75 label "c" , numbers between label "b" . the result would be

array(['a', 'a', 'a', 'b', ..., 'c'])

I tried things like arr[arr<0.25]='a' , but it will only work once since there will be strings comparing with float on the next command. I can achieve this by checking its type before comparison using a for loop, but it is complicated. Is there a straight forward way to achieve this?

Answer 1

NumPy arrays are homogeneous. You have to set type for label array

import numpy as np
arr=np.linspace(0,1,11)
lbl=np.empty((arr.shape), dtype=object)
lbl[arr<.25]='a'
lbl[(arr>=.25) & (arr <=.75)] = 'b'
lbl[arr>.75]='c'

print arr
print lbl

Output:

[ 0.   0.1  0.2  0.3  0.4  0.5  0.6  0.7  0.8  0.9  1. ]
['a' 'a' 'a' 'b' 'b' 'b' 'b' 'b' 'c' 'c' 'c']

Answer 2

For creating an array of three such groups, you could do something like this -

ID = (arr>0.75)*1 + (arr>=0.25)
select_arr = np.array(['a','b','c'])
out = select_arr[ID]

Sample run -

In [64]: arr # Modified from sample posted in question to include 0.75
Out[64]: 
array([ 0.  ,  0.1 ,  0.2 ,  0.3 ,  0.4 ,  0.5 ,  0.6 ,  0.7 ,  0.75,
        0.9 ,  1.  ])

In [65]: ID = (arr>0.75)*1 + (arr>=0.25)
    ...: select_arr = np.array(['a','b','c'])
    ...: out = select_arr[ID]
    ...: 

In [66]: out
Out[66]: 
array(['a', 'a', 'a', 'b', 'b', 'b', 'b', 'b', 'b', 'c', 'c'], 
      dtype='|S1')