public median(axis as Long, keepDims as Boolean = false, ignoreNaN as Boolean = false, where as Boolean[] = null, result as NDArray<Double> = null) as NDArray<Double>
Exact sample median by axes; O(largest slice) reusable scratch, source unchanged.
| Parameter | Description |
|---|---|
axis | Axis selection; negative indices are accepted. |
keepDims | Keep reduced dimensions with length one. |
result | Optional non-aliased, shape-compatible destination reused without allocation. |
ignoreNaN | Skip NaNs; otherwise propagate them. |
where | Optional flat Boolean mask, one entry per source element. |
Reduced Double tensor.
middle = data.median(axis=1)