scipy/cdf-complement
| Rule | cdf-complement |
|---|---|
| Pack | SciPy |
| Severity | warning |
| Fix | unsafe, with xpd check --fix --unsafe-fixes |
| Library versions | every version |
1 - dist.cdf(x) drops the digits of a small right tail and gives 0 once the cdf rounds to 1. Call dist.sf(x), which keeps them.
pvalue = 2 * (1 - stats.norm.cdf(np.abs(z)))pvalue = 2 * stats.norm.sf(np.abs(z))What the rule reports
Section titled “What the rule reports”The rule reports 1 - dist.cdf(...) and 1 - dist.sf(...) for a scipy.stats distribution
from a measured list. For stats.norm, 1 - cdf lost 3 or more digits above 4.25 and gave 0
above 8.3. lognorm, f, poisson, binom, nbinom and geom are not on the list, because
their sf lost digits too.
How to fix it
Section titled “How to fix it”xpd check --fix --unsafe-fixes applies the fixes of this rule. Check the result, because an unsafe fix can change what the code does.
The unsafe fix writes dist.sf(...), dist.cdf(...).
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
Subtracting $R.cdf from 1 drops the digits of a small right tail and gives 0 once the cdf rounds to 1. In the runs that lost up to all 16 digits, and $R.sf was never a digit worse. Write $R.sf($$$A).Browse the other rules of the SciPy pack.