NumPy | nanargmax method
Start your free 7-days trial now!
Numpy's nanargmax(~) method ignores all missing values (i.e. NaNs) returns the index that corresponds to the largest element in the array.
Parameters
1. a | array_like
The input array.
2. axis | int | optional
The axis along which to compute the method. For 2D arrays, if axis=0, then the method is performed column-wise, and if axis=1 then row-wise. If no axis is provided, then Numpy will deem your array as a flattened array.
Return value
If no axis is provided, then a scalar is returned. Otherwise, a Numpy array is returned.
Examples
One-dimensional arrays
x = np.array([np.NaN,5,1,3])np.nanargmax(x)
1
Here, 1 is returned because the largest value (i.e. 5) is located at index 1. In contrast, the np.argmax(x) method would return 0 since it considers NaN to be the largest.
Two-dimensional arrays
Suppose we have the following 2D Numpy array:
x = np.array([[np.NaN,4],[1,3]])x
array([[nan, 4.], [ 1., 3.]])
Max index of entire array
To obtain the index of the maximum value in the entire array, leave out the axis parameter:
np.nanargmax(x)
1
Max indices of every column
To obtain the index of the maximum values column-wise, set axis=0:
np.nanargmax(x, axis=0)
array([1, 0])
Here, we're going over each column of the matrix and computing the index of its largest value, while ignoring any missing values.
Max indices of every row
To obtain the index of the maximum values row-wise, set axis=1:
np.nanargmax(x, axis=1)
array([1, 1])
Here, we're going over each row of the matrix and computing the index of its largest value. while ignoring any missing values.
