res_df, k_labels_df = run_kmeans(X, y)
k_labels_df.to_csv(f"data/kmeans_labels_{data_key}.csv", index=False)
res_df.plot(
subplots=True,
style=".-",
title=f"KMeans performance vs n_clusters on {data_key} data",
)
plt.xlabel("n_clusters")
plt.savefig(f"output/kmeans_{data_key}.png")
plt.close()
if "credit" in data_key:
examine_credit_cluster(
X.values[:, TOP_FEATURES[:2]],
y,
title="True Label",
xylabel=DATA["credit"][0].columns[TOP_FEATURES[:2]],
fname=f"output/true_cluster_{data_key}.png",
)
examine_credit_cluster(
X.values[:, TOP_FEATURES[:2]],
k_labels_df[y.nunique()],
title="KMeans",
xylabel=DATA["credit"][0].columns[TOP_FEATURES[:2]],
fname=f"output/kmeans_cluster_{data_key}.png",
)
if data_key == "fashion":
plot_fashion_cluster(X,
k_labels_df[len(np.unique(y))],
fname="output/kmeans_cluster_fashion.png")
X_ica = ica.fit_transform(X)
kurt = kurtosis(X_ica)
plt.plot(kurt)
plt.xlabel("IC")
plt.ylabel("kurtosis")
plt.title(f"ICA kurtosis on {data_key} data")
plt.savefig(f"output/ica_kurtosis_{data_key}.png")
plt.close()
top_kurt_ind = (-kurt).argsort()
# IC
examine_credit_cluster(
X_ica[:, top_kurt_ind[:2]],
y,
title=f"ICA transformation on {data_key} data",
xylabel=["IC1", "IC2"],
fname=f"output/ica_ic_cluster_{data_key}.png",
)
if data_key == "fashion":
# IC
plot_fashion_cluster(
ica.components_[top_kurt_ind[:25], :],
range(25),
fname="output/ica_ic_fashion.png",
)
# reconstructed image
X_recon = ica.inverse_transform(X_ica)
plot_fashion_cluster(X_recon,
print(f"Running KMeans on {data_key} data")
for algo_key in DATA[data_key]:
print(f"..Running on {algo_key} transformed data")
X, y = DATA[data_key][algo_key]
res_df, k_labels_df = run_kmeans(X, y)
k_labels_df.to_csv(f"data/kmeans_labels_{algo_key}_{data_key}.csv",
index=False)
res_df.plot(
subplots=True,
style=".-",
title=
f"KMeans performance vs n_clusters on {algo_key} transformed {data_key} data",
)
plt.xlabel("n_clusters")
plt.savefig(f"output/kmeans_{algo_key}_{data_key}.png")
plt.close()
examine_credit_cluster(
X[:, :2],
y,
title=f"True label of {algo_key} transformed {data_key} data",
fname=f"output/true_cluster_{algo_key}_{data_key}.png",
)
examine_credit_cluster(
X[:, :2],
k_labels_df[len(np.unique(y))],
title=f"KMeans on {algo_key} transformed {data_key} data",
fname=f"output/kmeans_cluster_{algo_key}_{data_key}.png",
)
for data_key in DATA:
if data_key not in RUN_DATA:
continue
print(f"Running EM on {data_key} data")
X, y = DATA[data_key]
res_df, k_labels_df = run_em(X, y)
k_labels_df.to_csv(f"data/EM_labels_{data_key}.csv", index=False)
res_df.plot(
subplots=True,
style=".-",
title=f"EM performance vs n_clusters on {data_key} data",
)
plt.xlabel("n_clusters")
plt.savefig(f"output/EM_{data_key}.png")
plt.close()
if "credit" in data_key:
examine_credit_cluster(
X.values[:, TOP_FEATURES[:2]],
k_labels_df[y.nunique()],
title="EM",
xylabel=DATA["credit"][0].columns[TOP_FEATURES[:2]],
fname=f"output/EM_cluster_{data_key}.png",
)
if data_key == "fashion":
plot_fashion_cluster(X,
k_labels_df[len(np.unique(y))],
fname="output/EM_cluster_fashion.png")
index=NUM_CLUSTERS,
)
k_labels_df = pd.DataFrame(data=k_labels)
return res_df, k_labels_df
for data_key in DATA:
if data_key not in RUN_DATA:
continue
print(f"Running EM on {data_key} data")
for algo_key in DATA[data_key]:
print(f"..Running on {algo_key} transformed data")
X, y = DATA[data_key][algo_key]
res_df, k_labels_df = run_em(X, y)
k_labels_df.to_csv(f"data/EM_labels_{algo_key}_{data_key}.csv", index=False)
res_df.plot(
subplots=True,
style=".-",
title=f"EM performance vs n_clusters on {algo_key} transformed {data_key} data",
)
plt.xlabel("n_clusters")
plt.savefig(f"output/EM_{algo_key}_{data_key}.png")
plt.close()
examine_credit_cluster(
X[:, :2],
k_labels_df[len(np.unique(y))],
title=f"EM on {algo_key} transformed {data_key} data",
fname=f"output/EM_cluster_{algo_key}_{data_key}.png",
)
X_pca = pca.fit_transform(X)
eigenvalues = pca.explained_variance_
plt.plot(eigenvalues)
if data_key == "fashion":
plt.yscale("log")
plt.xlabel("PC")
plt.ylabel("eigenvalue")
plt.title(f"PCA eigenvalue on {data_key} data")
plt.savefig(f"output/pca_eigenval_{data_key}.png")
plt.close()
# PC
examine_credit_cluster(
X_pca[:, :2],
y,
title=f"PCA transformation on {data_key} data",
xylabel=["PC1", "PC2"],
fname=f"output/pca_pc_cluster_{data_key}.png",
)
if data_key == "fashion":
# PC image
plot_fashion_cluster(
pca.components_[:25, :], range(25), fname="output/pca_pc_fashion.png"
)
# reconstructed image
pca = PCA(n_components=0.95, whiten=True, random_state=0)
X_recon = pca.inverse_transform(pca.fit_transform(X))
print(f"...Keep {pca.n_components_} components for {data_key} data...")
plot_fashion_cluster(X_recon, y, fname="output/pca_reconstructed_fashion.png")
from sklearn.manifold import TSNE
from load_data import DATA
from examine_cluster import examine_credit_cluster
RUN_DATA = ["credit", "fashion"]
for data_key in DATA:
if data_key not in RUN_DATA:
continue
print(f"Running TSNE on {data_key} data")
X, y = DATA[data_key]
tsne = TSNE(n_components=3, random_state=0, n_jobs=-1)
X_tsne = tsne.fit_transform(X)
examine_credit_cluster(
X_tsne[:, :2],
y,
title=f"TSNE on {data_key} data",
xylabel=["", ""],
fname=f"output/tsne_cluster_{data_key}.png",
)