diff options
| author | Mike Jarvis <michael@jarvis.net> | 2021-06-08 07:41:59 -0400 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2021-06-08 14:41:59 +0300 |
| commit | 5c4aac16b54284a59bc5af34e731be7299cbb1d1 (patch) | |
| tree | e3310251db564f44e9b18e77402955c6a2c4632e /numpy/lib/polynomial.py | |
| parent | 8f1eff4d9b09741d1e989e8118d67af4de7990c5 (diff) | |
| download | numpy-5c4aac16b54284a59bc5af34e731be7299cbb1d1.tar.gz | |
DOC: Adjust polyfit doc to clarify the meaning of w (#18421)
* DOC: Adjust polyfit doc to clarify the meaning of w
cov='unscaled', in particular, had inconsistently referred to a weight
of 1/sigma**2, while the doc for w says it should be equal to 1/sigma.
This change clarifies w to comport with more typical meanings of
weights in weighted least squares, and makes clear that cov='unscaled'
is appropriate when the weight w**2 = 1/sigma**2.
See Issue #5261 for more discussion of the errors/confusion in
the previous doc string.
* Update doc text for w in all polynomial module fit functions
Co-authored-by: Stefan van der Walt <sjvdwalt@gmail.com>
Co-authored-by: Ross Barnowski <rossbar@berkeley.edu>
Diffstat (limited to 'numpy/lib/polynomial.py')
| -rw-r--r-- | numpy/lib/polynomial.py | 9 |
1 files changed, 6 insertions, 3 deletions
diff --git a/numpy/lib/polynomial.py b/numpy/lib/polynomial.py index 56fcce621..23021cafa 100644 --- a/numpy/lib/polynomial.py +++ b/numpy/lib/polynomial.py @@ -489,8 +489,11 @@ def polyfit(x, y, deg, rcond=None, full=False, w=None, cov=False): default) just the coefficients are returned, when True diagnostic information from the singular value decomposition is also returned. w : array_like, shape (M,), optional - Weights to apply to the y-coordinates of the sample points. For - gaussian uncertainties, use 1/sigma (not 1/sigma**2). + Weights. If not None, the weight ``w[i]`` applies to the unsquared + residual ``y[i] - y_hat[i]`` at ``x[i]``. Ideally the weights are + chosen so that the errors of the products ``w[i]*y[i]`` all have the + same variance. When using inverse-variance weighting, use + ``w[i] = 1/sigma(y[i])``. The default value is None. cov : bool or str, optional If given and not `False`, return not just the estimate but also its covariance matrix. By default, the covariance are scaled by @@ -498,7 +501,7 @@ def polyfit(x, y, deg, rcond=None, full=False, w=None, cov=False): to be unreliable except in a relative sense and everything is scaled such that the reduced chi2 is unity. This scaling is omitted if ``cov='unscaled'``, as is relevant for the case that the weights are - 1/sigma**2, with sigma known to be a reliable estimate of the + w = 1/sigma, with sigma known to be a reliable estimate of the uncertainty. Returns |