Ejemplo empresas en Reorganización - Regularización#
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.ticker as mtick
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from keras.models import Sequential
from keras.layers import Dense, InputLayer, Dropout
from keras.optimizers import Adam
from keras.optimizers import RMSprop
from keras.callbacks import EarlyStopping
from sklearn.metrics import classification_report, confusion_matrix, roc_auc_score, roc_curve
from sklearn.metrics import ConfusionMatrixDisplay, precision_score, precision_recall_curve, recall_score, accuracy_score, f1_score
# path = "BD empresas re organización.xlsx"
path = "BD empresas en re organización.xlsx"
xls = pd.ExcelFile(path)
df = pd.read_excel(path, sheet_name=xls.sheet_names[0])
df.head()
| Razón Social | Margen EBIT | Carga financiera | Margen neto | CxC | CxP | Solvencia | Apalancamiento | En Reorganización | |
|---|---|---|---|---|---|---|---|---|---|
| 0 | AACER SAS | 0.071690 | 0.000000 | 0.042876 | 0.104095 | 0.153192 | 1.877078 | 1.642505 | 0 |
| 1 | ABARROTES EL ROMPOY SAS | 0.017816 | 0.000000 | 0.010767 | 0.018414 | 0.000000 | 0.000000 | 0.865044 | 0 |
| 2 | ABASTECIMIENTOS INDUSTRIALES SAS | 0.144646 | 0.054226 | 0.059784 | 0.227215 | 0.025591 | 1.077412 | 1.272299 | 0 |
| 3 | ACME LEON PLASTICOS SAS | 0.004465 | 0.000000 | -0.013995 | 0.073186 | 0.127866 | 0.000000 | 1.391645 | 0 |
| 4 | ADVANCED PRODUCTS COLOMBIA SAS | 0.141829 | 0.050810 | 0.053776 | 0.398755 | 0.147678 | 0.675073 | 2.118774 | 0 |
df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 629 entries, 0 to 628
Data columns (total 9 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 Razón Social 629 non-null object
1 Margen EBIT 629 non-null float64
2 Carga financiera 629 non-null float64
3 Margen neto 629 non-null float64
4 CxC 629 non-null float64
5 CxP 629 non-null float64
6 Solvencia 629 non-null float64
7 Apalancamiento 629 non-null float64
8 En Reorganización 629 non-null int64
dtypes: float64(7), int64(1), object(1)
memory usage: 44.4+ KB
# Conteo absoluto
conteo_clases = df['En Reorganización'].value_counts()
# Porcentaje
porcentaje_clases = df['En Reorganización'].value_counts(normalize=True) * 100
# Mostrar conteo y porcentaje
print("Cantidad de empresas por clase:")
print(conteo_clases)
print("\nPorcentaje de empresas por clase:")
print(porcentaje_clases.round(2))
Cantidad de empresas por clase:
En Reorganización
1 342
0 287
Name: count, dtype: int64
Porcentaje de empresas por clase:
En Reorganización
1 54.37
0 45.63
Name: proportion, dtype: float64
Red Neuronal Artificial:#
df.columns
Index(['Razón Social', 'Margen EBIT', 'Carga financiera', 'Margen neto', 'CxC',
'CxP', 'Solvencia', 'Apalancamiento', 'En Reorganización'],
dtype='object')
# ------------------------
# Selección de variables
# ------------------------
variables_seleccionadas = ['Margen EBIT',
'Carga financiera',
'Margen neto',
'CxC',
'CxP',
'Solvencia',
'Apalancamiento']
# Variable objetivo
target = 'En Reorganización'
# ------------------------
# Preparar datos
# ------------------------
X = df[variables_seleccionadas]
y = df[target]
# Estandarizar variables
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)
# Dividir en entrenamiento y prueba (70%-30%)
X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.3, random_state=35, stratify=y)
X_train.shape, X_test.shape, y_train.shape, y_test.shape
((440, 7), (189, 7), (440,), (189,))
type(X_train), type(X_test), type(y_train), type(y_test)
(numpy.ndarray,
numpy.ndarray,
pandas.core.series.Series,
pandas.core.series.Series)
X_train.shape[1]
7
model = Sequential()
model.add(InputLayer(shape=(X_train.shape[1],)))
model.add(Dense(10, activation='relu'))
model.add(Dense(10, activation='relu'))
model.add(Dense(1, activation='sigmoid')) # 'sigmoid' porque es clasifiación binaria
model.compile(loss='binary_crossentropy', optimizer=Adam(learning_rate=0.001), metrics=['accuracy'])
history = model.fit(X_train, y_train,
validation_data=(X_test, y_test),
epochs=200,
batch_size=32,
verbose=0
)
# Graficar Loss y Val Loss:
plt.figure(figsize=(5, 5))
plt.plot(history.history["loss"], color="blue")
plt.plot(history.history["val_loss"], color="red")
plt.title("Loss vs Val Loss")
plt.legend(["loss", "val_loss"])
plt.show()
Regularización#
Dropout
Early Stopping
model = Sequential()
model.add(InputLayer(shape=(X_train.shape[1],)))
model.add(Dense(10, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(10, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='binary_crossentropy', optimizer=Adam(learning_rate=0.001), metrics=['accuracy'])
callback = EarlyStopping(monitor='val_loss', patience=10, restore_best_weights=False)
history = model.fit(X_train, y_train,
validation_data=(X_test, y_test),
epochs=200,
batch_size=32,
callbacks=[callback],
verbose=0
)
# Graficar Loss y Val Loss:
plt.figure(figsize=(5, 5))
plt.plot(history.history["loss"], color="blue")
plt.plot(history.history["val_loss"], color="red")
plt.title("Loss vs Val Loss")
plt.legend(["loss", "val_loss"])
plt.show()
y_prob_train = model.predict(X_train, verbose=0).flatten()
y_prob_test = model.predict(X_test, verbose=0).flatten()
# Calcular y_pred_train y y_pred_train con umbral de 0.5
umbral = 0.5
y_pred_train = np.where(y_prob_train >= umbral, 1, 0)
y_pred = np.where(y_prob_test >= umbral, 1, 0)
# ------------------------
# Evaluación del modelo
# ------------------------
cm_train = confusion_matrix(y_train, y_pred_train, labels=[0,1])
cm_df_train = pd.DataFrame(cm_train, index=["Real 0", "Real 1"], columns=["Predicho 0", "Predicho 1"])
plt.figure(figsize=(5.2,4.2))
sns.heatmap(cm_train, annot=True, fmt="d", cbar=True, linewidths=.5, cmap="coolwarm")
plt.title("Matriz de confusión - train")
plt.xlabel("Predicho"); plt.ylabel("Real")
plt.tight_layout()
plt.show()
cm = confusion_matrix(y_test, y_pred, labels=[0,1])
cm_df = pd.DataFrame(cm, index=["Real 0", "Real 1"], columns=["Predicho 0", "Predicho 1"])
plt.figure(figsize=(5.2,4.2))
sns.heatmap(cm_df, annot=True, fmt="d", cbar=True, linewidths=.5, cmap="coolwarm")
plt.title("Matriz de confusión - Test")
plt.xlabel("Predicho"); plt.ylabel("Real")
plt.tight_layout()
plt.show()
print("\n=== Reporte de Clasificación - train ===")
print(classification_report(y_train, y_pred_train))
print("\n=== Reporte de Clasificación - test ===")
print(classification_report(y_test, y_pred))
=== Reporte de Clasificación - train ===
precision recall f1-score support
0 0.72 0.90 0.80 201
1 0.89 0.71 0.79 239
accuracy 0.80 440
macro avg 0.81 0.81 0.80 440
weighted avg 0.82 0.80 0.80 440
=== Reporte de Clasificación - test ===
precision recall f1-score support
0 0.69 0.83 0.75 86
1 0.83 0.69 0.75 103
accuracy 0.75 189
macro avg 0.76 0.76 0.75 189
weighted avg 0.76 0.75 0.75 189
# DataFrame con probas y clase real
df_deciles = pd.DataFrame({'y_real': y_test, 'y_proba': y_prob_test})
# Crear deciles (1 = más alto riesgo, 10 = más bajo)
df_deciles['Decil'] = pd.qcut(df_deciles['y_proba'], 10, labels=False, duplicates='drop') + 1
df_deciles['Decil'] = 11 - df_deciles['Decil'] # invertir para que el decil 1 sea el de mayor riesgo
# Calcular tasa por decil
tabla_deciles = df_deciles.groupby('Decil').agg(
Total=('y_real','count'),
Positivos=('y_real','sum')
)
tabla_deciles['Tasa'] = tabla_deciles['Positivos'] / tabla_deciles['Total']
tabla_deciles['Lift'] = tabla_deciles['Tasa'] / df_deciles['y_real'].mean()
tabla_deciles['Captura_Acum'] = tabla_deciles['Positivos'].cumsum() / df_deciles['y_real'].sum()
print(f"Tasa de positivos reales en test: {df_deciles['y_real'].mean():.2f}")
print(tabla_deciles)
# --- 📊 Gráfico ---
plt.figure(figsize=(8,5))
plt.plot(tabla_deciles.index, tabla_deciles['Tasa'], marker='o', linestyle='-', color='blue')
plt.title("Tasa de positivos por decil")
plt.xlabel("Decil")
plt.ylabel("Tasa de clase 1")
plt.grid(True)
plt.show()
Tasa de positivos reales en test: 0.54
Total Positivos Tasa Lift Captura_Acum
Decil
1 19 19 1.000000 1.834951 0.184466
2 19 19 1.000000 1.834951 0.368932
3 19 17 0.894737 1.641799 0.533981
4 19 11 0.578947 1.062340 0.640777
5 18 9 0.500000 0.917476 0.728155
6 19 11 0.578947 1.062340 0.834951
7 19 3 0.157895 0.289729 0.864078
8 19 7 0.368421 0.676035 0.932039
9 19 5 0.263158 0.482882 0.980583
10 19 2 0.105263 0.193153 1.000000
