percentileofscore - rewrite?

Following ticket #560: http://projects.scipy.org/scipy/scipy/ticket/560

rewrote percentileofscore without using "histogram". The function now assigns a percentile by the relative location of the input relative to an array, when the array is sorted.

[stefanv, 2008-11-16 14:12:20]

Ariel, thanks for your work on this!  I hope to see your patch included in 0.7.

http://codereview.appspot.com/7913/diff/1/2
File scipy/stats/stats.py (right):

http://codereview.appspot.com/7913/diff/1/2#newcode1017
Line 1017: def percentileofscore(a, score):
Thanks for the rewrite.  The new implementation is much cleaner!  Please add
some tests to verify behaviour.

http://codereview.appspot.com/7913/diff/1/2#newcode1020
Line 1020: 
Let's rephrase this sentence for clarity, and add an extended description as
well.  Something like (please improve)

Percentile of a score.

The percentile of a score is the number of values smaller than or equal to the
score.  For example, if you score in the 80th percentile, it means that 80% of
the class had a mark that was the same as or lower than yours.

http://codereview.appspot.com/7913/diff/1/2#newcode1024
Line 1024: 
a : ndarray

http://codereview.appspot.com/7913/diff/1/2#newcode1026
Line 1026: 
score : float or int

http://codereview.appspot.com/7913/diff/1/2#newcode1033
Line 1033: 
Remove blank line.

Add a simpler example, e.g.:

>>> percentileofscore([20,80,100],80)

Woops -- bug!

http://codereview.appspot.com/7913/diff/1/2#newcode1035
Line 1035: >>> percentileofscore(a,101.0)
Spacing, see blow.

http://codereview.appspot.com/7913/diff/1/2#newcode1051
Line 1051: idx=[a==score]
Please refer to PEP8 for spacing guidelines:

a = np.sort(a)
a_len = np.array(range(len(a)))
idx = [a == score]

http://codereview.appspot.com/7913/diff/1/2#newcode1053
Line 1053: pct = (np.mean(a_len[idx])/(len(a)-1))*100.
Bug is in this line:

len(a)-1 should be used only if `score` was appended!

[arokem, 2008-11-22 06:34:29]

Stefan - I have changed the code according to your comments. Can you take a look
at it now? 

Cheers -- Ariel 


On 2008/11/16 14:12:20, Stefan wrote:
> Ariel, thanks for your work on this!  I hope to see your patch included in
0.7.
> 
> http://codereview.appspot.com/7913/diff/1/2
> File scipy/stats/stats.py (right):
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1017
> Line 1017: def percentileofscore(a, score):
> Thanks for the rewrite.  The new implementation is much cleaner!  Please add
> some tests to verify behaviour.
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1020
> Line 1020: 
> Let's rephrase this sentence for clarity, and add an extended description as
> well.  Something like (please improve)
> 
> Percentile of a score.
> 
> The percentile of a score is the number of values smaller than or equal to the
> score.  For example, if you score in the 80th percentile, it means that 80% of
> the class had a mark that was the same as or lower than yours.
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1024
> Line 1024: 
> a : ndarray
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1026
> Line 1026: 
> score : float or int
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1033
> Line 1033: 
> Remove blank line.
> 
> Add a simpler example, e.g.:
> 
> >>> percentileofscore([20,80,100],80)
> 
> Woops -- bug!
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1035
> Line 1035: >>> percentileofscore(a,101.0)
> Spacing, see blow.
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1051
> Line 1051: idx=[a==score]
> Please refer to PEP8 for spacing guidelines:
> 
> a = np.sort(a)
> a_len = np.array(range(len(a)))
> idx = [a == score]
> 
> http://codereview.appspot.com/7913/diff/1/2#newcode1053
> Line 1053: pct = (np.mean(a_len[idx])/(len(a)-1))*100.
> Bug is in this line:
> 
> len(a)-1 should be used only if `score` was appended!