A simple example of Kernel Density Estimation in NumPy.

For setup and testing, make sure to install pytest and setuptools. Install via pip:

pip install kde_numpy/

Run tests:

python -m pytest

Assuming the availability of training and test sets, we can use the KDE as follows:

# instantiate KDE
kde = get_KDE(batch_size=128, kernel_type='gaussian', eps=1e-8, bandwidth=0.2)

# set parameters
kde.fit(train_X)

# score test set
mlp = kde.mean_log_prob(test_X)