Adding tests directory into machine learning(using pytest)

[miladsade96]

Hi @cclauss , I would like to write tests into linear_discriminant_analysis functions using doctest, but doctest have not enough options and features to do this correctly ; therefore i decided to use pytest instead. we need to add a tests directory and write tests into it.
what is your opinion?!

[cclauss]

https://github.com/TheAlgorithms/Python/blob/master/strings/rabin_karp.py#L51-L55 uses a doctest to bootstrap the execution of a normal unittest. Would that technique work for your purposes?

[miladsade96]

I don't think so, one of the problems is that i can not handle multiple line generated list (expected output) in interpreter mode in docstring as follows :
for example:
>>> gaussian_distribution(5, 1, 20)

[6.288184753155463, 6.4494456086997705, 5.066335808938262, 4.235456349028368,
3.9078267848958586, 5.031334516831717, 3.977896829989127, 3.56317055489747,
5.199311976483754, 5.133374604658605, 5.546468300338232, 4.086029056264687,
5.005005283626573, 4.935258239627312, 3.494170998739258, 5.537997178661033,
5.320711100998849, 7.3891120432406865, 5.202969177309964, 4.855297691835079]

[cclauss]

>>> gaussian_distribution(5, 1, 20) # doctest: +NORMALIZE_WHITESPACE

We already do this a lot...
https://github.com/TheAlgorithms/Python/search?q=NORMALIZE_WHITESPACE

[miladsade96]

I didn't know this option, Thank you very much.

[miladsade96]

When i wrote a doctest into <predict_y_values> function , i got print keyword output as expected value instead of function return value.
Should i remove line below from that function?
print("Generated Discriminants: \n", results)
I require your assistance, @cclauss