Proyecto Final CoderHouse

# Importamos las librerias import pandas as pd import numpy as np import plotly.express as px import seaborn as sns import matplotlib.pyplot as plt

## Obtenemos el dataset del historial crediticio historial_credito = pd.read_csv('data/credit_record.csv') ## Obtenemos el dataset del registro de aplicación de credito registro_aplicacion = pd.read_csv('data/application_record.csv')

registro_aplicacion.head()

IDint64

CODE_GENDERobject

5008804

5008805

5008806

5008808

5008809

historial_credito.head()

IDint64

MONTHS_BALANCEint64

5001711

-1

5001711

-2

5001711

-3

5001712

## Info de los datasets print(historial_credito.shape) historial_credito.info()

(1048575, 3)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1048575 entries, 0 to 1048574
Data columns (total 3 columns):
 #   Column          Non-Null Count    Dtype 
---  ------          --------------    ----- 
 0   ID              1048575 non-null  int64 
 1   MONTHS_BALANCE  1048575 non-null  int64 
 2   STATUS          1048575 non-null  object
dtypes: int64(2), object(1)
memory usage: 24.0+ MB

print(registro_aplicacion.shape) registro_aplicacion.info()

(438557, 18)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 438557 entries, 0 to 438556
Data columns (total 18 columns):
 #   Column               Non-Null Count   Dtype  
---  ------               --------------   -----  
 0   ID                   438557 non-null  int64  
 1   CODE_GENDER          438557 non-null  object 
 2   FLAG_OWN_CAR         438557 non-null  object 
 3   FLAG_OWN_REALTY      438557 non-null  object 
 4   CNT_CHILDREN         438557 non-null  int64  
 5   AMT_INCOME_TOTAL     438557 non-null  float64
 6   NAME_INCOME_TYPE     438557 non-null  object 
 7   NAME_EDUCATION_TYPE  438557 non-null  object 
 8   NAME_FAMILY_STATUS   438557 non-null  object 
 9   NAME_HOUSING_TYPE    438557 non-null  object 
 10  DAYS_BIRTH           438557 non-null  int64  
 11  DAYS_EMPLOYED        438557 non-null  int64  
 12  FLAG_MOBIL           438557 non-null  int64  
 13  FLAG_WORK_PHONE      438557 non-null  int64  
 14  FLAG_PHONE           438557 non-null  int64  
 15  FLAG_EMAIL           438557 non-null  int64  
 16  OCCUPATION_TYPE      304354 non-null  object 
 17  CNT_FAM_MEMBERS      438557 non-null  float64
dtypes: float64(2), int64(8), object(8)
memory usage: 60.2+ MB

## Numero de datos nulos y su porcentaje en las columnas porcentaje_nulos = registro_aplicacion.isnull().mean().round(4) * 100 pd.merge(registro_aplicacion.isnull().sum().reset_index(), porcentaje_nulos.reset_index(), how='inner', on='index', suffixes=("_Valores Nulos", "_Porcentaje de valores nulos"))

indexobject

ID5.6%

CODE_GENDER5.6%

16 others88.9%

0_Valores Nulosint64

0 - 134203

CODE_GENDER

FLAG_OWN_CAR

FLAG_OWN_REALTY

CNT_CHILDREN

AMT_INCOME_TOTAL

NAME_INCOME_TYPE

NAME_EDUCATION_TYPE

NAME_FAMILY_STATUS

NAME_HOUSING_TYPE

DAYS_BIRTH

DAYS_EMPLOYED

FLAG_MOBIL

FLAG_WORK_PHONE

FLAG_PHONE

FLAG_EMAIL

OCCUPATION_TYPE

134203

CNT_FAM_MEMBERS

## Combinar ambos dataframes para hacer un concentrado por ID data = pd.merge(historial_credito, registro_aplicacion, how='inner', on='ID') data

IDint64

MONTHS_BALANCEint64

5008804

-1

5008804

-2

5008804

-3

5008804

-4

5008804

-5

5008804

-6

5008804

-7

5008804

-8

5008804

-9

## Eliminamos columnas que no vamos a ocupar data.drop(['NAME_INCOME_TYPE', 'FLAG_PHONE', 'FLAG_MOBIL', 'FLAG_EMAIL', 'FLAG_WORK_PHONE', 'OCCUPATION_TYPE'], axis=1, inplace=True)

## Nos quedamos con el último mes de actividad data = data[data['MONTHS_BALANCE'] == 0] data

IDint64

5008804 - 5150487

MONTHS_BALANCEint64

0 - 0

5008804

5008805

5008806

5008808

5008810

5008811

154

5008813

188

5008815

211

5008821

234

5008824

## Creamos la variable edad data['Edad'] = (data['DAYS_BIRTH'] / 365) * -1 ## Funcion lambda tiene_empleo = lambda x: True if x < 0 else False # Aplicamos la función a la nueva variable data['Empleo'] = data.DAYS_EMPLOYED.apply(tiene_empleo) ## Presentamos el DataFrame data

<ipython-input-12-735bc8430817>:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  data['Edad'] = (data['DAYS_BIRTH'] / 365) * -1
<ipython-input-12-735bc8430817>:8: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  data['Empleo'] = data.DAYS_EMPLOYED.apply(tiene_empleo)

IDint64

5008804 - 5150487

MONTHS_BALANCEint64

0 - 0

5008804

5008805

5008806

5008808

5008810

5008811

154

5008813

188

5008815

211

5008821

234

5008824

# Eliminamos las variables antiguas data.drop(['DAYS_BIRTH', 'DAYS_EMPLOYED'], axis=1, inplace=True)

<ipython-input-13-160f3f9f692b>:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  data.drop(['DAYS_BIRTH', 'DAYS_EMPLOYED'], axis=1, inplace=True)

# Checamos el tipo de dato de edad print(data.Edad.dtypes) # Lo pasamos a entero data['Edad'] = data['Edad'].astype(int) # Volvemos a checar el tipo de dato de edad print(data.Edad.dtypes)

float64
int64
<ipython-input-14-7a285713298a>:5: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  data['Edad'] = data['Edad'].astype(int)

# Vamos a renombrar la variable STATUS como target y vamos a reorganizar su contenido target = lambda x: 1 if x in ('C', 'X', '0') else 0 data['Target'] = data.STATUS.apply(target) data

<ipython-input-15-9f1714e3feac>:3: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  data['Target'] = data.STATUS.apply(target)

IDint64

5008804 - 5150487

MONTHS_BALANCEint64

0 - 0

5008804

5008805

5008806

5008808

5008810

5008811

154

5008813

188

5008815

211

5008821

234

5008824

# Vemos como esta la proporción de los buenos con los malos data.Target.value_counts()

# Eliminamos las columnas no necesarias data_procesada = data.drop(['MONTHS_BALANCE', 'STATUS', 'CODE_GENDER', 'NAME_FAMILY_STATUS', 'NAME_EDUCATION_TYPE', 'NAME_HOUSING_TYPE'], axis=1) data_procesada

IDint64

5008804 - 5150487

FLAG_OWN_CARobject

N62%

Y38%

5008804

5008805

5008806

5008808

5008810

5008811

154

5008813

188

5008815

211

5008821

234

5008824

# Ponemos el ID como index data_procesada.set_index('ID', inplace=True)

# Hacemos el cambio de variables booleanas a número data_procesada.FLAG_OWN_CAR = data_procesada.FLAG_OWN_CAR.replace({"Y": 1, "N": 0}) data_procesada.FLAG_OWN_REALTY = data_procesada.FLAG_OWN_REALTY.replace({"Y": 1, "N": 0}) data_procesada.Empleo = data_procesada.Empleo.replace({True: 1, False: 0})

data_procesada

FLAG_OWN_CARint64

0 - 1

FLAG_OWN_REALTYint64

0 - 1

5008804

5008805

5008806

5008808

5008810

5008811

5008813

5008815

5008821

5008824

data_procesada.describe().round()

FLAG_OWN_CARfloat64

FLAG_OWN_REALTYfloat64

count

24672

mean

std

min

25%

50%

75%

max

# Histogramas fig, axs = plt.subplots(ncols=4, nrows=2, figsize=(20,10)) index = 0 axs = axs.flatten() for k,v in data_procesada.items(): sns.distplot(v, ax=axs[index]) index += 1 plt.tight_layout(pad=0.4, w_pad=0.5, h_pad=5.0)

/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
/shared-libs/python3.9/py/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)

from scipy import stats fig, axs = plt.subplots(ncols=4, nrows=2, figsize=(20,10)) index = 0 axs = axs.flatten() for k,v in data_procesada.items(): sns.boxplot(y=k, data=data_procesada, ax=axs[index]) index += 1 plt.tight_layout(pad=0.4, w_pad=0.5, h_pad=5.0)

for k, v in data_procesada.items(): q1 = v.quantile(0.25) q3 = v.quantile(0.75) irq = q3 - q1 v_col = v[(v <= q1 - 1.5 * irq) | (v >= q3 + 1.5 * irq)] perc = np.shape(v_col)[0] * 100.0 / np.shape(data_procesada)[0] print("Variable %s outliers = %.2f%%" % (k, perc))

Variable FLAG_OWN_CAR outliers = 0.00%
Variable FLAG_OWN_REALTY outliers = 0.00%
Variable CNT_CHILDREN outliers = 1.41%
Variable AMT_INCOME_TOTAL outliers = 4.27%
Variable CNT_FAM_MEMBERS outliers = 1.34%
Variable Edad outliers = 0.00%
Variable Empleo outliers = 100.00%
Variable Target outliers = 100.00%

data_limpia = data_procesada[data_procesada.AMT_INCOME_TOTAL <= 387000] data_limpia = data_limpia[data_procesada.CNT_CHILDREN <= 2] data_limpia = data_limpia[data_procesada.CNT_FAM_MEMBERS <= 4]

<ipython-input-25-bf4e26441da1>:2: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  data_limpia = data_limpia[data_procesada.CNT_CHILDREN <= 2]
<ipython-input-25-bf4e26441da1>:3: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  data_limpia = data_limpia[data_procesada.CNT_FAM_MEMBERS <= 4]

# Comparamos las filas entre dataframes print(data_procesada.shape) print(data_limpia.shape)

(24672, 8)
(23409, 8)

# Checamos que ya no haya outliers variables = ['CNT_CHILDREN', 'AMT_INCOME_TOTAL', 'CNT_FAM_MEMBERS'] fig, axes = plt.subplots(1, len(variables), figsize=(15,6)) for ax, variable in zip(axes, variables): ax = sns.boxplot( y=variable, data=data_limpia, ax=ax) plt.tight_layout() plt.show()

#Resulta más intuitivo ver la matriz de correlación a través de un gráfico - heatmap de seaborn plt.figure(figsize=(20,10)) #Defino el tamaño del grafico ## Definimos la correlación sns.heatmap(data_limpia.corr(), annot=True)

fig = px.scatter(data_limpia, x="Edad", y="AMT_INCOME_TOTAL", color="Target") fig.show()

data_limpia

FLAG_OWN_CARint64

0 - 1

FLAG_OWN_REALTYint64

0 - 1

5008806

5008808

5008810

5008811

5008813

5008815

5008821

5008824

5008825

5008827

# Convertimos nuestra variable Target a número para poder hacer la regresión logistica y las demás métricas data_limpia.Target = data_limpia.Target.replace({"Bueno": 1, "Malo": 0}) X = data_limpia.drop("Target", axis=1) #Elimino de mi dataset la variable a predecir y = data_limpia.Target #Defino el Target

# (Si no esta instalada la libreria) !pip install imblearn

Collecting imblearn
  Downloading imblearn-0.0-py2.py3-none-any.whl (1.9 kB)
Collecting imbalanced-learn
  Downloading imbalanced_learn-0.9.0-py3-none-any.whl (199 kB)
     |████████████████████████████████| 199 kB 18.1 MB/s 
Requirement already satisfied: joblib>=0.11 in /shared-libs/python3.9/py/lib/python3.9/site-packages (from imbalanced-learn->imblearn) (1.1.0)
Requirement already satisfied: threadpoolctl>=2.0.0 in /shared-libs/python3.9/py/lib/python3.9/site-packages (from imbalanced-learn->imblearn) (3.1.0)
Requirement already satisfied: scipy>=1.1.0 in /shared-libs/python3.9/py/lib/python3.9/site-packages (from imbalanced-learn->imblearn) (1.8.0)
Requirement already satisfied: scikit-learn>=1.0.1 in /shared-libs/python3.9/py/lib/python3.9/site-packages (from imbalanced-learn->imblearn) (1.0.2)
Requirement already satisfied: numpy>=1.14.6 in /shared-libs/python3.9/py-core/lib/python3.9/site-packages (from imbalanced-learn->imblearn) (1.22.3)
Installing collected packages: imbalanced-learn, imblearn
Successfully installed imbalanced-learn-0.9.0 imblearn-0.0
WARNING: You are using pip version 21.2.3; however, version 22.0.4 is available.
You should consider upgrading via the '/root/venv/bin/python -m pip install --upgrade pip' command.

# importamos SMOTE para el balanceo de los datos from imblearn.over_sampling import SMOTE sm = SMOTE(random_state = 2)

# Balanceamos la información X, y= sm.fit_resample(X, y) # Vemos como esta la proporción de los buenos con los malos y.value_counts()

from sklearn.model_selection import train_test_split #Importamos la libreria para separar en train y test #Me quedo con 30% para test y 70% para train X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.30, random_state=42)

from sklearn.tree import DecisionTreeClassifier #Importamos el modelo arbol_de_decision = DecisionTreeClassifier(max_depth=2, random_state = 42) #Creamos el modelo

arbol_de_decision.fit(X_train, y_train)

y_train_pred = arbol_de_decision.predict(X_train) #Prediccion en Train y_test_pred = arbol_de_decision.predict(X_test) #Prediccion en Test

from sklearn.metrics import accuracy_score #Calculo el accuracy en Train train_accuracy = accuracy_score(y_train, y_train_pred) #Calculo el accuracy en Test test_accuracy_decisionTree = accuracy_score(y_test, y_test_pred) print('% de aciertos sobre el set de entrenamiento:', train_accuracy) print('% de aciertos sobre el set de evaluación:',test_accuracy_decisionTree)

% de aciertos sobre el set de entrenamiento: 0.6411326470679155
% de aciertos sobre el set de evaluación: 0.6436350522899387

from sklearn.metrics import confusion_matrix from sklearn.metrics import ConfusionMatrixDisplay cm = confusion_matrix(y_test, y_test_pred, labels= arbol_de_decision.classes_) disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=arbol_de_decision.classes_) disp.plot() plt.show()

X = data_limpia.drop("Target", axis=1) #Elimino de mi dataset la variable a predecir y = data_limpia.Target #Defino el Target

# Balanceamos la información X, y= sm.fit_resample(X, y) # Vemos como esta la proporción de los buenos con los malos y.value_counts()

#Separamos en train y test! (X_train, X_test, y_train, y_test) = train_test_split(X, y, stratify=y, test_size=0.30, random_state=10)

# Importamos el modelo from sklearn.ensemble import RandomForestClassifier #Creamos un random forest! model = RandomForestClassifier(random_state=10, max_features="log2") model.fit(X_train, y_train)

y_train_pred = model.predict(X_train) #Prediccion en Train y_test_pred = model.predict(X_test) #Prediccion en Test

#Calculo el accuracy en Train train_accuracy = accuracy_score(y_train, y_train_pred) #Calculo el accuracy en Test test_accuracy_randomForest = accuracy_score(y_test, y_test_pred) print('% de aciertos sobre el set de entrenamiento:', train_accuracy) print('% de aciertos sobre el set de evaluación:',test_accuracy_randomForest)

% de aciertos sobre el set de entrenamiento: 0.96343009057467
% de aciertos sobre el set de evaluación: 0.9542733501622791

cm = confusion_matrix(y_test, y_test_pred, labels= model.classes_) disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=model.classes_) disp.plot()

from sklearn.metrics import precision_score from sklearn.metrics import recall_score from sklearn.metrics import f1_score print("Recall:") # Calculo de recall en Train train_recall = recall_score(y_train, y_train_pred) # Calculo de recall en Test test_recall = recall_score(y_test, y_test_pred) print('% de recall sobre el set de entrenamiento:', train_recall) print('% de recall sobre el set de evaluación:', test_recall) print("Precisión:") # Calculo de precisión en Train train_precision = precision_score(y_train, y_train_pred) # Calculo de precisión en Test test_precision = precision_score(y_test, y_test_pred) print('% de precisión sobre el set de entrenamiento:', train_precision) print('% de precisión sobre el set de evaluación:', test_precision) print("F1 Score:") # Calculo de f1 score en Train train_f1_score = f1_score(y_train, y_train_pred) # Calculo de f1 score en Test test_f1_score = f1_score(y_test, y_test_pred) print('% de f1 score sobre el set de entrenamiento:', train_f1_score) print('% de f1_score sobre el set de evaluación:', test_f1_score)

Recall:
% de recall sobre el set de entrenamiento: 0.9488685544701373
% de recall sobre el set de evaluación: 0.939564402134718
Precisión:
% de precisión sobre el set de entrenamiento: 0.9773291727695345
% de precisión sobre el set de evaluación: 0.9680487442413435
F1 Score:
% de f1 score sobre el set de entrenamiento: 0.9628886030680428
% de f1_score sobre el set de evaluación: 0.9535939101156492

## Separamos las data X = data_limpia.drop("Target", axis=1) #Elimino de mi dataset la variable a predecir y = data_limpia.Target #Defino el Target

X, y= sm.fit_resample(X, y) # Vemos como esta la proporción de los buenos con los malos y.value_counts()

#Separamos en train y test! (X_train, X_test, y_train, y_test) = train_test_split(X, y, stratify=y, test_size=0.30, random_state=11)

# Importamos la libreria del modelo from sklearn.linear_model import LogisticRegression # Creamos una instancia de la Regresión Logística regresion_logistica = LogisticRegression() # Entrenamos la regresión logística con los datos de entrenamiento regresion_logistica.fit(X_train,y_train)

y_train_pred = regresion_logistica.predict(X_train) #Prediccion en Train y_test_pred = regresion_logistica.predict(X_test) #Prediccion en Test

#Calculo el accuracy en Train train_accuracy = accuracy_score(y_train, y_train_pred) #Calculo el accuracy en Test test_accuracy_regresion = accuracy_score(y_test, y_test_pred) print('% de aciertos sobre el set de entrenamiento:', train_accuracy) print('% de aciertos sobre el set de evaluación:',test_accuracy_regresion)

% de aciertos sobre el set de entrenamiento: 0.5015920121178398
% de aciertos sobre el set de evaluación: 0.4980887125856473

print("Recall:") # Calculo de recall en Train train_recall = recall_score(y_train, y_train_pred) # Calculo de recall en Test test_recall = recall_score(y_test, y_test_pred) print('% de recall sobre el set de entrenamiento:', train_recall) print('% de recall sobre el set de evaluación:', test_recall) print("Precisión:") # Calculo de precisión en Train train_precision = precision_score(y_train, y_train_pred) # Calculo de precisión en Test test_precision = precision_score(y_test, y_test_pred) print('% de precisión sobre el set de entrenamiento:', train_precision) print('% de precisión sobre el set de evaluación:', test_precision) print("F1 Score:") # Calculo de f1 score en Train train_f1_score = f1_score(y_train, y_train_pred) # Calculo de f1 score en Test test_f1_score = f1_score(y_test, y_test_pred) print('% de f1 score sobre el set de entrenamiento:', train_f1_score) print('% de f1_score sobre el set de evaluación:', test_f1_score)

Recall:
% de recall sobre el set de entrenamiento: 0.6349078768393718
% de recall sobre el set de evaluación: 0.630318765325256
Precisión:
% de precisión sobre el set de entrenamiento: 0.5012446917557476
% de precisión sobre el set de evaluación: 0.4985169974903034
F1 Score:
% de f1 score sobre el set de entrenamiento: 0.5602138512315539
% de f1_score sobre el set de evaluación: 0.55672335817568

# Desplegamos los valores de cada modelo para compararlos print('% de aciertos sobre el set de evaluación del Árbol de decisión:',test_accuracy_decisionTree) print('% de aciertos sobre el set de evaluación del Random Forest:',test_accuracy_randomForest) print('% de aciertos sobre el set de evaluación de la Regresión Logistica:',test_accuracy_regresion)

% de aciertos sobre el set de evaluación del Árbol de decisión: 0.6436350522899387
% de aciertos sobre el set de evaluación del Random Forest: 0.9542733501622791
% de aciertos sobre el set de evaluación de la Regresión Logistica: 0.4980887125856473

from sklearn.model_selection import LeaveOneOut from sklearn.model_selection import cross_val_score from numpy import mean from numpy import std

X = data_limpia.drop("Target", axis=1) #Elimino de mi dataset la variable a predecir y = data_limpia.Target #Defino el Target

# Balanceo los datos X, y= sm.fit_resample(X, y) # Vemos como esta la proporción de los buenos con los malos y.value_counts()

#Creamos el modelo cv = LeaveOneOut()

# Decision Tree model = RandomForestClassifier(random_state=10, max_features="log2")

# Evaluacion del Modelo # scores = cross_val_score(model, X, y, scoring='accuracy', cv=cv, n_jobs=-1)

# print('Accuracy: %.3f (%.3f)' % (mean(scores), std(scores)))

import scipy as sp from sklearn.model_selection import RandomizedSearchCV from sklearn.metrics import confusion_matrix, accuracy_score from sklearn.metrics import classification_report

X = data_limpia.drop("Target", axis=1) #Elimino de mi dataset la variable a predecir y = data_limpia.Target #Defino el Target

# Balanceamos la data X, y= sm.fit_resample(X, y) # Vemos como esta la proporción de los buenos con los malos y.value_counts()

#Separamos en train y test! (X_train, X_test, y_train, y_test) = train_test_split(X, y, stratify=y, test_size=0.30, random_state=10)

#Creamos nuestro objeto Decision Tree random_forest = RandomForestClassifier()

# Grilla para Random Search param_dist = {'criterion': ['gini', 'entropy'], 'max_depth': sp.stats.randint(1, 20), 'min_samples_split': sp.stats.randint(1, 20), 'random_state': sp.stats.randint(1, 20), 'min_samples_leaf': sp.stats.randint(1, 20), 'max_features': ['auto', 'sqrt', 'log2']} #Aplicamos la grilla al modelo model = RandomizedSearchCV(random_forest, param_dist, n_iter=100, random_state=0, cv=5)

#Entrenamos DecisionTree con la grilla definida arriba y CV con tamaño de Fold=5 model.fit(X_train, y_train)

/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/model_selection/_validation.py:372: FitFailedWarning:


30 fits failed out of a total of 500.
The score on these train-test partitions for these parameters will be set to nan.
If these failures are not expected, you can try to debug them by setting error_score='raise'.

Below are more details about the failures:
--------------------------------------------------------------------------------
30 fits failed with the following error:
Traceback (most recent call last):
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/model_selection/_validation.py", line 680, in _fit_and_score
    estimator.fit(X_train, y_train, **fit_params)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/ensemble/_forest.py", line 450, in fit
    trees = Parallel(
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/joblib/parallel.py", line 1043, in __call__
    if self.dispatch_one_batch(iterator):
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/joblib/parallel.py", line 861, in dispatch_one_batch
    self._dispatch(tasks)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/joblib/parallel.py", line 779, in _dispatch
    job = self._backend.apply_async(batch, callback=cb)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/joblib/_parallel_backends.py", line 208, in apply_async
    result = ImmediateResult(func)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/joblib/_parallel_backends.py", line 572, in __init__
    self.results = batch()
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/joblib/parallel.py", line 262, in __call__
    return [func(*args, **kwargs)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/joblib/parallel.py", line 262, in <listcomp>
    return [func(*args, **kwargs)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/utils/fixes.py", line 216, in __call__
    return self.function(*args, **kwargs)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/ensemble/_forest.py", line 185, in _parallel_build_trees
    tree.fit(X, y, sample_weight=curr_sample_weight, check_input=False)
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/tree/_classes.py", line 937, in fit
    super().fit(
  File "/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/tree/_classes.py", line 250, in fit
    raise ValueError(
ValueError: min_samples_split must be an integer greater than 1 or a float in (0.0, 1.0]; got the integer 1


/shared-libs/python3.9/py/lib/python3.9/site-packages/sklearn/model_selection/_search.py:969: UserWarning:

One or more of the test scores are non-finite: [0.94151292 0.86339601 0.91805017 0.89838946 0.86293226 0.64323466
        nan 0.76568062 0.66329739 0.92636557 0.84181858 0.9156388
 0.87718321 0.93347565 0.85764626 0.92988965 0.811277   0.73866325
        nan 0.90481937 0.76849368 0.93492847 0.89090855 0.66329739
 0.72741025 0.86188127 0.9266748  0.89829667 0.90182076 0.78413541
 0.90092432 0.82330179 0.91495881 0.80630016 0.88506593 0.92639657
 0.87600841 0.75551013 0.68308085 0.86491054 0.92451084 0.76342373
 0.75931246 0.94352217 0.91464964 0.88228384 0.89208328 0.92642741
 0.64416213        nan        nan 0.78694838 0.94240934 0.92315077
 0.89359794 0.93007523 0.72920314 0.68657391 0.76153829 0.91789551
 0.91427871 0.85356557 0.90475758 0.91718443 0.83931474 0.77684001
 0.77449071 0.90689052 0.90132609 0.70119613 0.84209724 0.77201757
 0.63739221 0.84231339        nan 0.88930104 0.8680948  0.88985749
 0.81925244 0.78447524 0.89786391 0.78781394 0.71000667 0.82008723
 0.88549881 0.92308888 0.65532137 0.76966831 0.8987914         nan
 0.72586538 0.82135438 0.82614607 0.8957     0.85359642 0.93041519
 0.87641037 0.80614546 0.75643748 0.7876597 ]

print("Mejores parametros: "+str(model.best_params_)) print("Mejor Score: "+str(model.best_score_)+'\n')

Mejores parametros: {'criterion': 'entropy', 'max_depth': 19, 'max_features': 'sqrt', 'min_samples_leaf': 3, 'min_samples_split': 6, 'random_state': 6}
Mejor Score: 0.9435221743552594

#Analizamos qué obtuvimos scores = pd.DataFrame(model.cv_results_) scores

mean_fit_timefloat64

0.06512484550476075 - 2.744015407562256

std_fit_timefloat64

0.0027582693767645365 - 0.4861242494309885

1.741157627

0.04761143871

1.570005083

0.03744789727

1.578596067

0.05041297753

1.765538073

0.06826824218

1.35671196

0.01747660497

0.6017141342

0.01189156825

0.06520042419

0.002758269377

0.929206562

0.039771927

0.5702732086

0.01169563289

1.616813755

0.07309009584

#Prediccion prediction = model.predict(X_test)

#Accuracy print('Exactitud:', accuracy_score(y_test, prediction))

Exactitud: 0.9509556437071763

# Importamos la libreria del modelo from sklearn.ensemble import AdaBoostClassifier X = data_limpia.drop("Target", axis=1) #Elimino de mi dataset la variable a predecir y = data_limpia.Target #Defino el Target

# Balanceamos la data X, y= sm.fit_resample(X, y) # Vemos como esta la proporción de los buenos con los malos y.value_counts()

# Separación en Train y Test X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3) # 70% training y 30% test

# Creamos el modelo AdaBoostClassifier ada = AdaBoostClassifier(n_estimators=50, learning_rate=1) # Se fitea el modelo ada.fit(X_train, y_train)

# Predicción en Train y_train_pred = ada.predict(X_train) # Predicción en Test y_test_pred = ada.predict(X_test)

# Calculo de accuracy en Train train_accuracy = accuracy_score(y_train, y_train_pred) # Calculo de accuracy en Test test_accuracy = accuracy_score(y_test, y_test_pred) print('% de aciertos sobre el set de entrenamiento:', train_accuracy) print('% de aciertos sobre el set de evaluación:', test_accuracy)

% de aciertos sobre el set de entrenamiento: 0.7924510804043402
% de aciertos sobre el set de evaluación: 0.7886043995672557

cm = confusion_matrix(y_test, y_test_pred, labels= ada.classes_) disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=ada.classes_) disp.plot() plt.show()

Recall:
% de recall sobre el set de entrenamiento: 0.8537569026493764
% de recall sobre el set de evaluación: 0.8509298998569385
Precisión:
% de precisión sobre el set de entrenamiento: 0.7595076447535464
% de precisión sobre el set de evaluación: 0.7589638892433329
F1 Score:
% de f1 score sobre el set de entrenamiento: 0.8038791844365251
% de f1_score sobre el set de evaluación: 0.8023200917245565

# Instalamos la libreria si no la tenemos !pip install xgboost

Collecting xgboost
  Downloading xgboost-1.5.2-py3-none-manylinux2014_x86_64.whl (173.6 MB)
     |████████████████████████████████| 173.6 MB 43.5 MB/s 
Requirement already satisfied: scipy in /shared-libs/python3.9/py/lib/python3.9/site-packages (from xgboost) (1.8.0)
Requirement already satisfied: numpy in /shared-libs/python3.9/py-core/lib/python3.9/site-packages (from xgboost) (1.22.3)
Installing collected packages: xgboost
Successfully installed xgboost-1.5.2
WARNING: You are using pip version 21.2.3; however, version 22.0.4 is available.
You should consider upgrading via the '/root/venv/bin/python -m pip install --upgrade pip' command.

# Importamos la libreria del modelo import xgboost as xgb X = data_limpia.drop("Target", axis=1) #Elimino de mi dataset la variable a predecir y = data_limpia.Target #Defino el Target

/root/venv/lib/python3.9/site-packages/xgboost/compat.py:36: FutureWarning:

pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

# Balanceamos la data X, y= sm.fit_resample(X, y) # Vemos como esta la proporción de los buenos con los malos y.value_counts()

# Separación en Train y Test X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3) # 70% training y 30% test

clf_xgb = xgb.XGBClassifier(objective='binary:logistic', n_estimators=40, seed=123)

clf_xgb.fit(X_train, y_train)

/root/venv/lib/python3.9/site-packages/xgboost/sklearn.py:1224: UserWarning:

The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

/root/venv/lib/python3.9/site-packages/xgboost/data.py:262: FutureWarning:

pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

[03:08:06] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.

# Predicción en Train y_train_pred = clf_xgb.predict(X_train) # Predicción en Test y_test_pred = clf_xgb.predict(X_test)

% de aciertos sobre el set de entrenamiento: 0.9264583140127979
% de aciertos sobre el set de evaluación: 0.9191489361702128

cm = confusion_matrix(y_test, y_test_pred, labels= clf_xgb.classes_) disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=clf_xgb.classes_) disp.plot() plt.show()

Recall:
% de recall sobre el set de entrenamiento: 0.9371944805395706
% de recall sobre el set de evaluación: 0.9297437374028218
Precisión:
% de precisión sobre el set de entrenamiento: 0.9173834039975772
% de precisión sobre el set de evaluación: 0.9107319136934142
F1 Score:
% de f1 score sobre el set de entrenamiento: 0.9271831287686326
% de f1_score sobre el set de evaluación: 0.9201396309752796