K-Means

app.py

@@ -4,11 +4,11 @@ import numpy as np
 from bokeh.plotting import figure
 from bokeh.models import ColumnDataSource, DataTable, TableColumn, CustomJS, Select, Button, HoverTool, LinearColorMapper, ColorBar, FuncTickFormatter, FixedTicker
 from bokeh.layouts import column
-from bokeh.palettes import Reds9, Blues9, Oranges9, Purples9, Greys9, BuGn9, Greens9, RdYlGn11, linear_palette
+from bokeh.palettes import Reds9, Blues9, Oranges9, Purples9, Greys9, BuGn9, Greens9, RdYlGn11, linear_palette, Category10
 from sklearn.decomposition import PCA
 from sklearn.manifold import TSNE, trustworthiness
 from sklearn.metrics import pairwise_distances, silhouette_score
-from sklearn.cluster import DBSCAN
+from sklearn.cluster import DBSCAN, KMeans
 from sklearn.preprocessing import MinMaxScaler
 from sklearn.pipeline import Pipeline
 from sklearn.base import BaseEstimator, TransformerMixin
@@ -614,6 +614,15 @@ def compute_global_regression(df_combined, embedding_cols, tsne_params, df_f1, r
         silhouette_vals.append(silhouette_score(reduced_real, df_combined[real_mask]['label']))
 
         silhouette = np.max(silhouette_vals)
+    
+    inertias = []
+    K = range(1, 20)
+
+    for k in K:
+        kmeans = KMeans(n_clusters=k, random_state=42)
+        kmeans.fit(reduced_real)
+        inertias.append(kmeans.inertia_)
+
 
     dfs_reduced, unique_subsets = split_versions(df_combined, reduced)
     
@@ -689,7 +698,8 @@ def compute_global_regression(df_combined, embedding_cols, tsne_params, df_f1, r
         "explained_variance": explained_variance,
         "trustworthiness": trust,
         "continuity": cont,
-        "silhouette": silhouette
+        "silhouette": silhouette,
+        "inertias": inertias,
     }
     
     if reduction_method == "PCA":
@@ -828,6 +838,26 @@ def run_model(model_name):
     st.write(f"Continuity: {result['continuity']:.2f}")
     st.write(f"Silhouette Score: {result['silhouette']:.2f}")
 
+    st.subheader("Elbow Method (KMeans)")
+    if "inertias" in result and len(result["inertias"]) > 0:
+        K = np.arange(1, len(result["inertias"]) + 1)
+        inertias = np.array(result["inertias"])
+
+        source = ColumnDataSource(data=dict(K=K, inertias=inertias))
+        p = figure(title="Método del Codo",
+                x_axis_label='Número de Clusters (k)',
+                y_axis_label='Inercia (Within-Cluster Sum of Squares)',
+                width=600, height=400,
+                background_fill_color="#fafafa")
+
+        p.line('K', 'inertias', source=source, line_width=3, color=Category10[10][0])
+        p.circle('K', 'inertias', source=source, size=8, color=Category10[10][1], legend_label="Inertia")
+
+        p.legend.location = "top_right"
+        p.grid.visible = True
+
+        st.bokeh_chart(p, use_container_width=True)
+
     # # Mostrar los plots de loadings si se usó PCA (para el conjunto combinado)
     # if reduction_method == "PCA" and result.get("pca_model") is not None:
     #     # pca_model = result["pca_model"]