bit-lsi / men2017 Goto Github PK

Lines generally shouldn't go over 79. For the case of triple-quoted docstrings, it's easy to just split these over multiple lines.

Put these imports at the top of the file

There are a lot of places where you're getting sequential elements from a list. Use unpacking to make this code more readable:

At best, this can look like:

q_p,  q_m = predictionListStats

If there are more elements in predictionListStats, then you have a couple options:

Explicit assignment

q_p,  q_m = predictionListStats[0], predictionListStats[1] 

Unpacking sliced list

q_p,  q_m = predictionListStats[0:2]

Splat Unpacking

q_p,  q_m, *_ = predictionListStats

All functions could use unit tests

You can use vectorized operations in numpy to operate over entire arrays, then list unpack.

Also, remember that <- isn't a valid assingment in Python and the R people also finally decided to stop doing this too (it's only in legacy code or from legacy coders)

f_n_pp,  f_n_pm, f_n_mp, f_n_mm = np.floor(twoByTwoContingencyTable[0:4])

R and Python don't have the same documentation styles, so this has to be slightly re-written.

Python-style documentation also allows for embedding of mathtype and latex, so the yucky math looks beautiful

There must be a way to compute the log of the factorial directly to avoid dealing with ridiculous big numbers

At least, this code should be refactored since the pattern of sum log factorial is repeated many times throughout this repo