Modelado

Matriz de confusión

KNN

mc_knn = confusion_matrix(y, pred_knn)
mc_knn

array([[882, 425],
       [427, 856]])

Árbol

mc_arbol = confusion_matrix(y, pred_arbol)
mc_arbol

array([[993, 314],
       [312, 971]])

Bosques

mc_bosques = confusion_matrix(y, pred_bosques)
mc_bosques

array([[1097,  210],
       [ 224, 1059]])

Potenciación

mc_ada = confusion_matrix(y, pred_ada)
mc_ada

array([[1097,  210],
       [ 230, 1053]])

SVM

mc_svm = confusion_matrix(y, pred_svm)
mc_svm

array([[1098,  209],
       [ 205, 1078]])

Precisiones

def Errores(y_true, y_predicted):
  MC = confusion_matrix(y_true, y_predicted)
  PG = np.round(np.sum(MC.diagonal()) / np.sum(MC) * 100, 2)
  PC = np.round(MC.diagonal()/np.sum(MC, axis = 1) * 100, 2)
  df = pd.DataFrame({'PG' : [PG], 'PLOCAL' : [PC[0]], 'PVISITA' : [PC[1]]})
  return df

error_knn     = Errores(y, pred_knn)
error_svm     = Errores(y, pred_svm)
error_arbol   = Errores(y, pred_arbol)
error_bosques = Errores(y, pred_bosques)
error_ada     = Errores(y, pred_ada)

errores = pd.concat([
  error_knn, error_svm, error_arbol, error_bosques, error_ada], ignore_index = True)

errores.index = ['KNN', 'SVM', 'Árbol', 'Bosques', 'Potenciación']
errores

	PG	PLOCAL	PVISITA
KNN	67.10	67.48	66.72
SVM	84.02	84.01	84.02
Árbol	75.83	75.98	75.68
Bosques	83.24	83.93	82.54
Potenciación	83.01	83.93	82.07

fig = go.Figure()
no_plot = fig.add_trace(go.Bar(x = errores.index, y = errores.PG, name = "Global"))
no_plot = fig.add_trace(go.Bar(x = errores.index, y = errores.PLOCAL, name = "Local"))
no_plot = fig.add_trace(go.Bar(x = errores.index, y = errores.PVISITA, name = "Visita"))
fig.update_layout(title = "Precisiones")
fig.show()