Update pages/CGMM_Trial_1.py
Browse files- pages/CGMM_Trial_1.py +10 -10
pages/CGMM_Trial_1.py
CHANGED
|
@@ -220,11 +220,11 @@ def plot_glm_shap_importance(
|
|
| 220 |
#-------------------
|
| 221 |
# DATASETS
|
| 222 |
#-------------------
|
| 223 |
-
df1=pd.read_csv('.
|
| 224 |
-
df2=pd.read_csv('.
|
| 225 |
-
df1_synthetic=pd.read_csv('.
|
| 226 |
df1_synthetic = df1_synthetic.drop(columns=["Unnamed: 0"])
|
| 227 |
-
df2_synthetic=pd.read_csv('.
|
| 228 |
df2_synthetic = df2_synthetic.drop(columns=["Unnamed: 0"])
|
| 229 |
|
| 230 |
|
|
@@ -361,8 +361,8 @@ st.dataframe(styled_df)
|
|
| 361 |
st.subheader('PCA comparison')
|
| 362 |
st.write('Assesses similarity in global variance structure and major latent components.')
|
| 363 |
# Load the saved models
|
| 364 |
-
scaler = load('.
|
| 365 |
-
pca = load('.
|
| 366 |
|
| 367 |
def pca_visual_comparison_3d_with_saved_model(real_df, synthetic_df, scaler, pca, title_suffix=""):
|
| 368 |
assert set(real_df.columns) == set(synthetic_df.columns), "Datasets must have the same columns."
|
|
@@ -404,7 +404,7 @@ pca_visual_comparison_3d_with_saved_model(X_train, df1_synthetic, scaler, pca)
|
|
| 404 |
|
| 405 |
st.subheader('UMAP comparison')
|
| 406 |
st.write('Examines nonlinear manifold structure and clustering behavior.')
|
| 407 |
-
img = mpimg.imread('.
|
| 408 |
fig=plt.figure(figsize=(10, 8))
|
| 409 |
plt.imshow(img)
|
| 410 |
plt.axis('off')
|
|
@@ -653,8 +653,8 @@ st.dataframe(styled_df)
|
|
| 653 |
st.subheader('PCA comparison')
|
| 654 |
st.write('Assesses similarity in global variance structure and major latent components.')
|
| 655 |
# Load the saved models
|
| 656 |
-
scaler = load('.
|
| 657 |
-
pca = load('.
|
| 658 |
|
| 659 |
def pca_visual_comparison_3d_with_saved_model(real_df, synthetic_df, scaler, pca, title_suffix=""):
|
| 660 |
assert set(real_df.columns) == set(synthetic_df.columns), "Datasets must have the same columns."
|
|
@@ -689,7 +689,7 @@ pca_visual_comparison_3d_with_saved_model(X_train, df2_synthetic, scaler, pca)
|
|
| 689 |
# ### UMAP Analysis
|
| 690 |
st.subheader('UMAP comparison')
|
| 691 |
st.write('Examines nonlinear manifold structure and clustering behavior.')
|
| 692 |
-
img = mpimg.imread('.
|
| 693 |
fig=plt.figure(figsize=(10, 8))
|
| 694 |
plt.imshow(img)
|
| 695 |
plt.axis('off')
|
|
|
|
| 220 |
#-------------------
|
| 221 |
# DATASETS
|
| 222 |
#-------------------
|
| 223 |
+
df1=pd.read_csv('./data/ausprivauto0405.csv')
|
| 224 |
+
df2=pd.read_csv('./data/swmotorcycle.csv')
|
| 225 |
+
df1_synthetic=pd.read_csv('./CGMM/d1_cgmm_80_encod.csv')
|
| 226 |
df1_synthetic = df1_synthetic.drop(columns=["Unnamed: 0"])
|
| 227 |
+
df2_synthetic=pd.read_csv('./CGMM/d2_cgmm_80_encod.csv')
|
| 228 |
df2_synthetic = df2_synthetic.drop(columns=["Unnamed: 0"])
|
| 229 |
|
| 230 |
|
|
|
|
| 361 |
st.subheader('PCA comparison')
|
| 362 |
st.write('Assesses similarity in global variance structure and major latent components.')
|
| 363 |
# Load the saved models
|
| 364 |
+
scaler = load('./CGMM/scaler_pca_model_d1_cgmm_80.pkl')
|
| 365 |
+
pca = load('./CGMM/pca_model_d1_cgmm_80.pkl')
|
| 366 |
|
| 367 |
def pca_visual_comparison_3d_with_saved_model(real_df, synthetic_df, scaler, pca, title_suffix=""):
|
| 368 |
assert set(real_df.columns) == set(synthetic_df.columns), "Datasets must have the same columns."
|
|
|
|
| 404 |
|
| 405 |
st.subheader('UMAP comparison')
|
| 406 |
st.write('Examines nonlinear manifold structure and clustering behavior.')
|
| 407 |
+
img = mpimg.imread('./CGMM/umap_d1_80.png')
|
| 408 |
fig=plt.figure(figsize=(10, 8))
|
| 409 |
plt.imshow(img)
|
| 410 |
plt.axis('off')
|
|
|
|
| 653 |
st.subheader('PCA comparison')
|
| 654 |
st.write('Assesses similarity in global variance structure and major latent components.')
|
| 655 |
# Load the saved models
|
| 656 |
+
scaler = load('./CGMM/scaler_pca_model_d2_cgmm_80.pkl')
|
| 657 |
+
pca = load('./CGMM/pca_model_d2_cgmm_80.pkl')
|
| 658 |
|
| 659 |
def pca_visual_comparison_3d_with_saved_model(real_df, synthetic_df, scaler, pca, title_suffix=""):
|
| 660 |
assert set(real_df.columns) == set(synthetic_df.columns), "Datasets must have the same columns."
|
|
|
|
| 689 |
# ### UMAP Analysis
|
| 690 |
st.subheader('UMAP comparison')
|
| 691 |
st.write('Examines nonlinear manifold structure and clustering behavior.')
|
| 692 |
+
img = mpimg.imread('./CGMM/umap_d2_80.png')
|
| 693 |
fig=plt.figure(figsize=(10, 8))
|
| 694 |
plt.imshow(img)
|
| 695 |
plt.axis('off')
|