numpy.nanpercentile() in Python
numpy.nanpercentile() function compute the nth percentile of the given data (array elements) along the specified axis and ignores nan values. Example:
import numpy as np
a = np.array([10, 20, np.nan, 40])
res = np.nanpercentile(a, 50)
print(res)
Output
20.0
Explanation: The 50th percentile (median) is calculated over [10, 20, 40], ignoring the NaN.
Syntax
numpy.nanpercentile(a, q, axis=None, out=None, overwrite_input=False, interpolation='linear', keepdims=<no value>, method='linear')
Parameters:
Parameter | Description |
|---|---|
a | Input array. |
q | Percentile or sequence of percentiles (between 0 and 100). |
axis | Axis along which percentiles are computed. Default is None (entire array). |
out | Alternative output array. |
overwrite_input | If True, allows modifying input array (for memory efficiency). |
interpolation | [Deprecated] Use method instead. |
keepdims | If True, retains reduced dimensions. |
method | Specifies interpolation method (e.g., 'linear', 'nearest', 'lower', 'higher', 'midpoint'). Default: 'linear'. |
Returns: A scalar or ndarray of percentiles, depending on q and axis.
Examples
Example 1: Computing multiple percentiles
import numpy as np
a = np.array([5, 15, np.nan, 25])
res = np.nanpercentile(a, [25, 50, 75])
print(res)
Output
[10. 15. 20.]
Explanation: Calculates the 25th, 50th, and 75th percentiles of the non-NaN values.
Example 2: Applying along a specific axis
import numpy as np
a = np.array([[10, 20, np.nan], [30, np.nan, 50]])
res = np.nanpercentile(a, 50, axis=0)
print(res)
Output
[20. 20. 50.]
Explanation: The median of each column is computed by ignoring NaN values.
- Column 0: median(10, 30) = 20
- Column 1: median(20) = 20
- Column 2: median(50) = 50.
Example 3: Using different interpolation method
import numpy as np
a = np.array([1, 2, 3, 4, 5, np.nan])
res = np.nanpercentile(a, 40, method='nearest')
print(res)
Output
3.0
Explanation: With method='nearest', the function picks the value nearest to the 40th percentile position.
