feature/make_zn_scores_efficient

[Ksteeds]

def func_executor(df: pd.DataFrame, neutral: str, n: int):
"""
Function used to clean up the repetitive code in the below methods. Will produce the
evolving neutral level.

:param <pd.DataFrame> df:
:param <str> neutral:
:param <int> n: number of dates the neutral level is computed over.
"""

if neutral == "mean":
    return np.array([df.iloc[0:(i + 1), :].stack().mean() for i in range(n)])
else:
    return np.array([df.iloc[0:(i + 1), :].stack().median() for i in range(n)])

[Ksteeds]

Tied to the issue #342.

[Ksteeds]

Time taken for functional executor, median: 13.972623825073242
Overall time taken for make_zn_scores(): 24.274497985839844

[Ksteeds]

Time taken for functional executor, mean: 9.492730140686035
Overall time taken for make_zn_scores(): 19.86984610557556

Definitely scope to improve the above computational speed of a rolling mean.

[Ksteeds]

Following upgrades median algorithm takes: Time taken: 11.616146087646484.
Following upgrades mean algorithm takes: Time taken: 12.103456974029541

[Ksteeds]

def iis_std_panel(dfx: pd.DataFrame, min_obs: int, sequential: bool = True,
iis: bool = True):
"""
Function designed to compute the standard deviations but accounts for in-sampling
period. The in-sampling standard deviation will be a fixed value.
"""
Aim to improve the performance speed of this subroutine.