Reputation: 101
I have a set of values ranging from 3-27, that have 20 finite values:
A = [(0,21),(1,12),(2,15),(3,3),(4,21),(5,15),(6,27),(7,21),(8,9),(9,27),(10,12),(11,9),(12,12),(13,3),(14,9),(15,12),(16,6),(17,3),(18,9),(19,15)]
I would like to learn how to create a numpy array that would have 9 bins, each one having a range of -1 and +1 integer for the given tuples[1], the multiples of 3 from the range of 3-27 (but this should be interchangeable to any other combination of integers and range). In the end, I'd like to be able to create a matrix that looks something like this:
[[0,0,0,0,0,0,1,0,0],
[0,0,0,1,0,0,0,0,0],
[0,0,0,0,1,0,0,0,0],
[1,0,0,0,0,0,0,0,0],
[0,0,0,0,0,0,1,0,0],
[0,0,0,0,1,0,0,0,0],
[0,0,0,0,0,0,0,0,1],
....]
I was reading on how numpy has (num, bins) = histogram(x, bins=None, range=None)
but I'm not quite sure how to go by that.
I was thinking that I would have to iterate through 'A' to get the unique values('a') and then do the range by (a-1,a+1), to get the number of bins I would just to len(unique_values)
. But then I'm lost. Can anyone guide me?
Upvotes: 2
Views: 96
Reputation: 221574
Here's one way with np.searchsorted
/np.digitize
-
bins = np.arange(3,28,3)
ar = np.asarray(A)[:,1] # or np.array([i[1] for i in A])
ids = np.searchsorted(bins, ar) # or np.digitize(ar,bins)-1
out = (ids[:,None] == np.arange(9)).astype(int)
The last step to get the final output could be replaced by array-initialization -
out = np.zeros((len(ids), 9),dtype=int)
out[np.arange(len(ids)), ids] = 1
If the first element in the tuples were not in sequence, we might want to use those to index into rows -
out[np.asarray(A)[:,0], ids] = 1
Sample run -
In [205]: A
Out[205]:
[(0, 21),
(1, 12),
(2, 15),
(3, 3),
(4, 21),
(5, 15),
(6, 27),
(7, 21),
(8, 9),
(9, 27),
(10, 12),
(11, 9),
(12, 12),
(13, 3),
(14, 9),
(15, 12),
(16, 6),
(17, 3),
(18, 9),
(19, 15)]
In [206]: out[:7] # first 7 rows of output
Out[206]:
array([[0, 0, 0, 0, 0, 0, 1, 0, 0],
[0, 0, 0, 1, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 1]])
Upvotes: 2