Skip to content

Fixed error in KDE calculation #9

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
MaroGol wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fixed error in KDE calculation #9

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 10 additions & 2 deletions boom/data/prepare_splits_10k.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,10 +66,18 @@ def prepare_splits(
# Extract the hof values
hof_values = np.array(hof_values).reshape(-1, 1)

density_KDE = KernelDensity(kernel="gaussian", bandwidth="scott").fit(
# sklearn's "scott" assumes sigma=1, so we compute it manually:
# h = n^(-1/(d+4)) * sigma

n_samples_density, n_features_density = density_values.shape
scott_bw_density = n_samples_density ** (-1.0 / (n_features_density + 4)) * np.std(density_values)
density_KDE = KernelDensity(kernel="gaussian", bandwidth=scott_bw_density).fit(
density_values
)
hof_KDE = KernelDensity(kernel="gaussian", bandwidth="scott").fit(hof_values)

n_samples_hof, n_features_hof = hof_values.shape
scott_bw_hof = n_samples_hof ** (-1.0 / (n_features_hof + 4)) * np.std(hof_values)
hof_KDE = KernelDensity(kernel="gaussian", bandwidth=scott_bw_hof).fit(hof_values)

density_kde_scores = density_KDE.score_samples(density_values)
hof_kde_scores = hof_KDE.score_samples(hof_values)
Expand Down
7 changes: 6 additions & 1 deletion boom/data/prepare_splits_qm9.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -168,7 +168,12 @@ def prepare_splits_qm9(
property_values = np.array(property_values).reshape(-1, 1).astype(np.float64)

print("Starting Kernel Density Estimation for " + property_name)
property_KDE = KernelDensity(kernel="gaussian", bandwidth="scott").fit(
# sklearn's "scott" assumes sigma=1, so we compute it manually:
# h = n^(-1/(d+4)) * sigma

n_samples, n_features = property_values.shape
scott_bandwidth = n_samples ** (-1.0 / (n_features + 4)) * np.std(property_values)
property_KDE = KernelDensity(kernel="gaussian", bandwidth=scott_bandwidth).fit(
property_values
)
print("Kernel Density Estimation Done!")
Expand Down