Saving NaiveKDE-objects with pickle

[sbieringer]

Hi all,

Thanks for the package, it has been a great help! I am using version 1.1.0.
Whenever I try to save a NaiveKDE object with pickle, e.g.

import KDEpy
import pickle
import numpy as np

x = np.random.randn(1000)
kde = KDEpy.NaiveKDE('gaussian')
kde.fit(x)

with open('./kde.pt', 'wb')as f:
    pickle.dump(kde, f)

I get an error of the sort

      7 kde.fit(x)
      9 with open('./kde.pt', 'wb')as f:
---> 10     pickle.dump(kde, f)

PicklingError: Can't pickle <function gaussian at 0x7f517bca1430>: it's not the same object as KDEpy.kernel_funcs.gaussian

This could be easily resolved by changing lines 328-338 to names not coinciding with the function names, e.g.

gaussian_k = Kernel(gaussian, var=1, support=np.inf)
exp_k = Kernel(exponential, var=2, support=np.inf)
box_k = Kernel(box, var=1 / 3, support=1)
tri_k = Kernel(tri, var=1 / 6, support=1)
epa_k = Kernel(epanechnikov, var=1 / 5, support=1)
biweight_k = Kernel(biweight, var=1 / 7, support=1)
triweight_k = Kernel(triweight, var=1 / 9, support=1)
tricube_k = Kernel(tricube, var=35 / 243, support=1)
cosine_k = Kernel(cosine, var=(1 - (8 / np.pi**2)), support=1)
logistic_k = Kernel(logistic, var=(np.pi**2 / 3), support=np.inf)
sigmoid_k = Kernel(sigmoid, var=(np.pi**2 / 4), support=np.inf)

_kernel_functions = {
    "gaussian": gaussian_k,
    "exponential": exp_k,
    "box": box_k,
    "tri": tri_k,
    "epa": epa_k,
    "biweight": biweight_k,
    "triweight": triweight_k,
    "tricube": tricube_k,
    "cosine": cosine_k,
    # 'logistic': logistic_k,
    # 'sigmoid': sigmoid_k
}

[tommyod]

Feel free to add a test and open a pull request on this :) I would merge it if it's as easy as you say!