Assume a numpy array x. How can I do x[x>value] = max(value1, x[x>value])?
In words, if element x[i] > value, then x[i] = max(value1, x[i]).
Thanks.
EDIT
Using numpy.maximum instead of numpy.max solved the problem. When I had numpy.max I was getting the error message TypeError: only integer scalar arrays can be converted to a scalar index
Make an array:
In [361]: x = np.random.randint(0,10,10)
In [362]: x
Out[362]: array([7, 8, 4, 8, 1, 2, 6, 6, 3, 9])
Identify the values to be replaced:
In [363]: mask = x>5
In [364]: mask
Out[364]:
array([ True, True, False, True, False, False, True, True, False,
True])
In [365]: x[mask]
Out[365]: array([7, 8, 8, 6, 6, 9])
The replacement values:
In [368]: np.maximum(7, x[mask])
Out[368]: array([7, 8, 8, 7, 7, 9])
As long as the both sides have the same number of terms (shape actually):
In [369]: x[mask] = np.maximum(7, x[mask])
In [370]: x
Out[370]: array([7, 8, 4, 8, 1, 2, 7, 7, 3, 9])
Since this is actually just changing the values between 5 and 7, we could use:
In [378]: x = np.array([7, 8, 4, 8, 1, 2, 6, 6, 3, 9])
In [379]: mask = (x>5) & (x<7)
In [380]: mask
Out[380]:
array([False, False, False, False, False, False, True, True, False,
False])
In [381]: x[mask]
Out[381]: array([6, 6])
In [382]: x[mask] = 7
You can use np.where and single combined condition:
np.where((x>value)&(x<value1), value1, x)
or equivalently:
x[(x>value) & (x<value1)] = value1
Example:
x = np.arange(20).reshape(4,5)
value=5
value1=10
#output:
array([[ 0, 1, 2, 3, 4],
[ 5, 10, 10, 10, 10],
[10, 11, 12, 13, 14],
[15, 16, 17, 18, 19]])
