Loan Eligibility Analysis

Tentang Dataset

Dataset berikut digunakan untuk memprediksi apakah calon peminjam memenuhi syarat untuk disetujui peminjamannya berdasarkan wilayah, keuangan, dan faktor yang berhubungan dengan kredit lainnta. Dataset ini cocok untuk klasifikasi menggunakan machine learning.

Dataset diperoleh dari repository Kaggle pada link berikut

https://www.kaggle.com/datasets/avineshprabhakaran/loan-eligibility-prediction

Dataset memiliki cakupan data sebagai berikut

Column	Description	Example
Customer_ID	Unique identifier for each loan applicant	569
Gender	Gender of the applicant	Male / Female
Married	Marital status of the applicant	Yes / No
Dependents	Number of dependents	0, 1, 2, 3
Education	Education level of the applicant	Graduate / Not Graduate
Self_Employed	Whether the applicant is self-employed	Yes / No
Applicant_Income	Applicant’s monthly income	5000
Coapplicant_Income	Coapplicant’s monthly income	1500
Loan_Amount	Loan amount requested (in thousands)	128
Loan_Amount_Term	Loan repayment term (in months)	360
Credit_History	Credit history meets lending criteria	(1 = Yes, 0 = No) 1
Property_Area	Type of property area	Urban / Semiurban / Rural
Loan_Status	Loan approved or not (target variable)	Y / N

Import the library

import kagglehub
import os

import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.ticker import MultipleLocator
import matplotlib as mpl
import seaborn as sns

Unduh Dataset

path = kagglehub.dataset_download("avineshprabhakaran/loan-eligibility-prediction")
filename = os.listdir(path)[0]

fullpath = "/".join([path,filename])
fullpath

Downloading to /root/.cache/kagglehub/datasets/avineshprabhakaran/loan-eligibility-prediction/3.archive...

100%|██████████| 7.39k/7.39k [00:00<00:00, 13.1MB/s]

Extracting files...

'/root/.cache/kagglehub/datasets/avineshprabhakaran/loan-eligibility-prediction/versions/3/Loan Eligibility Prediction.csv'

Set the default plot style

def DefaultPlotStyle():
  plt.rcParams['lines.linewidth'] = 2
  plt.rcParams['lines.linestyle'] = '-'
  plt.rcParams['figure.figsize'] = [12,10]
  plt.rcParams['font.size'] = 12

DefaultPlotStyle()

Membaca Dataset

df = pd.read_csv(fullpath)
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 614 entries, 0 to 613
Data columns (total 13 columns):
 #   Column              Non-Null Count  Dtype
---  ------              --------------  -----
 0   Customer_ID         614 non-null    int64
 1   Gender              614 non-null    object
 2   Married             614 non-null    object
 3   Dependents          614 non-null    int64
 4   Education           614 non-null    object
 5   Self_Employed       614 non-null    object
 6   Applicant_Income    614 non-null    int64
 7   Coapplicant_Income  614 non-null    float64
 8   Loan_Amount         614 non-null    int64
 9   Loan_Amount_Term    614 non-null    int64
 10  Credit_History      614 non-null    int64
 11  Property_Area       614 non-null    object
 12  Loan_Status         614 non-null    object
dtypes: float64(1), int64(6), object(6)
memory usage: 62.5+ KB

Pengecekan Dataset

# Checking missing value
df.isnull().sum().reset_index().T

	0	1	2	3	4	5	6	7	8	9	10	11	12
index	Customer_ID	Gender	Married	Dependents	Education	Self_Employed	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Property_Area	Loan_Status
0	0	0	0	0	0	0	0	0	0	0	0	0	0

Tidak ada data atau nilai yang kosong dalam dataset ini

# Checking duplicate data
any(df.duplicated())

False

# Checking duplicate customer id
df['Customer_ID'].value_counts().reset_index().sort_values(by='count', ascending=False).head()

	Customer_ID	count
613	271	1
0	606	1
1	569	1
2	15	1
3	95	1

Tidak ada data yang duplikat dalam dataset ini

Tampilan Dataset

df.head()

	Customer_ID	Gender	Married	Dependents	Education	Self_Employed	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Property_Area	Loan_Status
0	569	Female	No	0	Graduate	No	2378	0.0	9	360	1	Urban	N
1	15	Male	Yes	2	Graduate	No	1299	1086.0	17	120	1	Urban	Y
2	95	Male	No	0	Not Graduate	No	3620	0.0	25	120	1	Semiurban	Y
3	134	Male	Yes	0	Graduate	Yes	3459	0.0	25	120	1	Semiurban	Y
4	556	Male	Yes	1	Graduate	No	5468	1032.0	26	360	1	Semiurban	Y

Analisis Parameter

# Creating object feature for Credit History
mapper = lambda x : "YES" if x == 1 else "NO"
df['Credit_History_Obj'] = df['Credit_History'].map(mapper)
df.head()

	Customer_ID	Gender	Married	Dependents	Education	Self_Employed	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Property_Area	Loan_Status	Credit_History_Obj
0	569	Female	No	0	Graduate	No	2378	0.0	9	360	1	Urban	N	YES
1	15	Male	Yes	2	Graduate	No	1299	1086.0	17	120	1	Urban	Y	YES
2	95	Male	No	0	Not Graduate	No	3620	0.0	25	120	1	Semiurban	Y	YES
3	134	Male	Yes	0	Graduate	Yes	3459	0.0	25	120	1	Semiurban	Y	YES
4	556	Male	Yes	1	Graduate	No	5468	1032.0	26	360	1	Semiurban	Y	YES

# Selecting only object features
object_list = df.select_dtypes(include=['object']).columns
object_list

Index(['Gender', 'Married', 'Education', 'Self_Employed', 'Property_Area',
       'Loan_Status', 'Credit_History_Obj'],
      dtype='object')

# Plotting object features
ndata = df.shape[0]
nobject = len(object_list)
n_cols = 3
n_rows = (nobject // n_cols) + 1

fig, ax = plt.subplots(nrows=n_rows, ncols=n_cols)
ax_flat = ax.flatten()

for i, ax in enumerate(ax_flat):
  if i < nobject:
    df_obj = df[object_list[i]].value_counts().reset_index()
    df_obj['percentage'] = (df_obj['count']/df_obj['count'].sum()) * 100
    ax = ax_flat[i]
    ax.pie(df_obj['percentage'], labels=df_obj[object_list[i]], autopct="%.1f%%")
    ax.set_title(object_list[i].replace("_", " "))
  else:
    ax.axis('off')

plt.tight_layout()

# Calculating the total income from the applicant income and co-applicant income
df['Total_Income'] = df['Applicant_Income'] + df['Coapplicant_Income']
df.head()

	Customer_ID	Gender	Married	Dependents	Education	Self_Employed	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Property_Area	Loan_Status	Credit_History_Obj	Total_Income
0	569	Female	No	0	Graduate	No	2378	0.0	9	360	1	Urban	N	YES	2378.0
1	15	Male	Yes	2	Graduate	No	1299	1086.0	17	120	1	Urban	Y	YES	2385.0
2	95	Male	No	0	Not Graduate	No	3620	0.0	25	120	1	Semiurban	Y	YES	3620.0
3	134	Male	Yes	0	Graduate	Yes	3459	0.0	25	120	1	Semiurban	Y	YES	3459.0
4	556	Male	Yes	1	Graduate	No	5468	1032.0	26	360	1	Semiurban	Y	YES	6500.0

fig, ax = plt.subplots(figsize=(12,6))
sns.countplot(data=df, x='Gender', hue='Loan_Status', ax=ax)
for container in ax.containers:
  ax.bar_label(container, label_type='edge')
# ax.set_ylim([0,200])
ax.set_ylabel('Number of Applicants')
ax.yaxis.set_minor_locator(MultipleLocator(25))
plt.grid(which='both')
plt.legend(title='Loan Status')
ax.set_axisbelow(True)

Berdasarkan jenis kelamin, peminjaman yang berhasil dilakukan secara mayoritas oleh pelanggan laki-kali, hanya kurang dari 100 pelanggan wanita yang berhasil mendapatkan pinjaman.

fig, ax = plt.subplots(figsize=(12,6))
sns.countplot(data=df, x='Education',hue='Loan_Status', ax=ax)
for container in ax.containers:
  ax.bar_label(container, label_type='edge')
# ax.set_ylim([0,200])
ax.set_ylabel('Number of Applicants')
ax.yaxis.set_minor_locator(MultipleLocator(25))
plt.grid(which='both')
plt.legend(title='Loan Status')
ax.set_axisbelow(True)

Berdasarkan status pendidikan, pelanggan yang lulus kuliah lebih berpeluang menerima pinjaman dibandingkan dengan pelanggan yang tidak lulus kuliah.

fig, ax = plt.subplots(figsize=(12,6))
sns.countplot(data=df, x='Self_Employed', hue='Loan_Status', ax=ax)
for container in ax.containers:
  ax.bar_label(container, label_type='edge')
# ax.set_ylim([0,200])
ax.set_ylabel('Number of Applicants')
ax.yaxis.set_minor_locator(MultipleLocator(25))
plt.grid(which='both')
plt.legend(title='Loan Status')
ax.set_axisbelow(True)

Berdasarkan status pekerjaan, pelanggan dengan status bukan wiraswasta berpeluang mendapatkan pinjaman.

fig, ax = plt.subplots(figsize=(12,6))
sns.countplot(data=df, x='Property_Area', hue='Loan_Status', ax=ax)
for container in ax.containers:
  ax.bar_label(container, label_type='edge')
# ax.set_ylim([0,200])
ax.set_ylabel('Number of Applicants')
ax.yaxis.set_minor_locator(MultipleLocator(25))
plt.grid(which='both')
ax.set_axisbelow(True)

Berdasarkan wilayah tempat tinggal, peminjaman banyak diajukan oleh pelanggan yang tinggal di wilayah semiurban, diikuti dengan pelanggan yang tinggal di wilayah urban dan pedesaan.

fig, ax = plt.subplots(figsize=(12,6))
sns.countplot(data=df, x='Dependents', hue='Loan_Status', ax=ax)
for container in ax.containers:
  ax.bar_label(container, label_type='edge')
# ax.set_ylim([0,200])
ax.set_ylabel('Number of Applicants')
ax.yaxis.set_minor_locator(MultipleLocator(25))
plt.grid(which='both')
ax.set_axisbelow(True)

Mayoritas peminjaman yang diajukan dan diterima berasal dari pelanggan yang tidak memiliki tanggungan.

fig, ax = plt.subplots(figsize=(12,6))
sns.countplot(data=df, x='Married', hue='Loan_Status', ax=ax)
for container in ax.containers:
  ax.bar_label(container, label_type='edge')
# ax.set_ylim([0,200])
ax.set_ylabel('Number of Applicants')
ax.yaxis.set_minor_locator(MultipleLocator(25))
plt.grid(which='both')
ax.set_axisbelow(True)

Berdasarkan status pernikahan, pelanggan dengan status sudah menikah berpeluang menerima pinjaman.

fig, ax = plt.subplots(figsize=(12,6))
sns.countplot(data=df, x='Credit_History_Obj', hue='Loan_Status', ax=ax)
for container in ax.containers:
  ax.bar_label(container, label_type='edge')
# ax.set_ylim([0,200])
ax.set_ylabel('Number of Applicants')
ax.yaxis.set_minor_locator(MultipleLocator(25))
plt.grid(which='both')
ax.set_axisbelow(True)

Berdasarkan rekam jejak peminjaman, pelanggan yang pernah melakukan peminjaman berpeluang mendapatkan pinjaman kembali.

# Selecting only numeric data
numeric_list = df.select_dtypes(exclude=['object']).columns
numeric_list = numeric_list.delete(0)
numeric_list

Index(['Dependents', 'Applicant_Income', 'Coapplicant_Income', 'Loan_Amount',
       'Loan_Amount_Term', 'Credit_History', 'Total_Income'],
      dtype='object')

df_numeric = df.select_dtypes(exclude=['object'])
df_numeric.head()

	Customer_ID	Dependents	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Total_Income
0	569	0	2378	0.0	9	360	1	2378.0
1	15	2	1299	1086.0	17	120	1	2385.0
2	95	0	3620	0.0	25	120	1	3620.0
3	134	0	3459	0.0	25	120	1	3459.0
4	556	1	5468	1032.0	26	360	1	6500.0

df_numeric.drop('Customer_ID', axis=1, inplace=True)
df_numeric.head()

	Dependents	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Total_Income
0	0	2378	0.0	9	360	1	2378.0
1	2	1299	1086.0	17	120	1	2385.0
2	0	3620	0.0	25	120	1	3620.0
3	0	3459	0.0	25	120	1	3459.0
4	1	5468	1032.0	26	360	1	6500.0

df.describe()

	Customer_ID	Dependents	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Total_Income
count	614.000000	614.000000	614.000000	614.000000	614.000000	614.000000	614.000000	614.000000
mean	307.500000	0.856678	5403.459283	1621.245798	142.022801	338.892508	0.850163	7024.705081
std	177.390811	1.216651	6109.041673	2926.248369	87.083089	69.716355	0.357203	6458.663872
min	1.000000	0.000000	150.000000	0.000000	9.000000	12.000000	0.000000	1442.000000
25%	154.250000	0.000000	2877.500000	0.000000	98.000000	360.000000	1.000000	4166.000000
50%	307.500000	0.000000	3812.500000	1188.500000	125.000000	360.000000	1.000000	5416.500000
75%	460.750000	2.000000	5795.000000	2297.250000	164.750000	360.000000	1.000000	7521.750000
max	614.000000	4.000000	81000.000000	41667.000000	700.000000	480.000000	1.000000	81000.000000

number_Coapplicant_Income_zero = df[df['Coapplicant_Income'] == 0].shape[0]
number_Coapplicant_Income_nonzero = df.shape[0] - number_Coapplicant_Income_zero

percentage_Coapplicant_Income_zero = (number_Coapplicant_Income_zero/df.shape[0])*100
percentage_Coapplicant_Income_nonzero = (number_Coapplicant_Income_nonzero/df.shape[0])*100

fig, ax = plt.subplots(figsize=(5,5))
ax.pie(x= [percentage_Coapplicant_Income_zero, percentage_Coapplicant_Income_nonzero],
       labels=['Do not have income', 'Have income'],
       autopct="%.1f%%");
ax.set_title('Coapplicant Income')

Text(0.5, 1.0, 'Coapplicant Income')

# Encoding the object data to numeric
df_object = df.select_dtypes(include=['object'])
df_object.head()

	Gender	Married	Education	Self_Employed	Property_Area	Loan_Status	Credit_History_Obj
0	Female	No	Graduate	No	Urban	N	YES
1	Male	Yes	Graduate	No	Urban	Y	YES
2	Male	No	Not Graduate	No	Semiurban	Y	YES
3	Male	Yes	Graduate	Yes	Semiurban	Y	YES
4	Male	Yes	Graduate	No	Semiurban	Y	YES

gender_encoder = lambda x : 1 if x == 'Male' else 0
married_encoder = lambda x : 1 if x == 'Yes' else 0
education_encoder = lambda x : 1 if x == 'Graduate' else 0
self_employed_encoder = lambda x : 1 if x == 'Yes' else 0
property_area_encoder = lambda x : 1 if x == 'Urban' else 2 if x == 'Semiurban' else 3
loan_status_encoder = lambda x : 1 if x == 'Y' else 0

mappers = [gender_encoder, married_encoder, education_encoder, self_employed_encoder, property_area_encoder, loan_status_encoder]

df_object_names = df_object.columns.to_list()
df_object_names.remove('Credit_History_Obj')
df_object_names

['Gender',
 'Married',
 'Education',
 'Self_Employed',
 'Property_Area',
 'Loan_Status']

df_duplicate = df.copy(deep=True)

for i, obj in enumerate(df_object_names):
  df_duplicate[obj] = df_duplicate[obj].map(mappers[i])

df_duplicate.drop(['Customer_ID','Credit_History_Obj'], axis=1, inplace=True)

df_duplicate.head()

	Gender	Married	Dependents	Education	Self_Employed	Applicant_Income	Coapplicant_Income	Loan_Amount	Loan_Amount_Term	Credit_History	Property_Area	Loan_Status	Total_Income
0	0	0	0	1	0	2378	0.0	9	360	1	1	0	2378.0
1	1	1	2	1	0	1299	1086.0	17	120	1	1	1	2385.0
2	1	0	0	0	0	3620	0.0	25	120	1	2	1	3620.0
3	1	1	0	1	1	3459	0.0	25	120	1	2	1	3459.0
4	1	1	1	1	0	5468	1032.0	26	360	1	2	1	6500.0

Applying Logistic Regression

from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix, accuracy_score, classification_report, roc_curve, roc_auc_score
from sklearn.preprocessing import StandardScaler
import numpy as np

target = df_duplicate['Loan_Status']
features = df_duplicate.drop('Loan_Status', axis=1)

# Splitting the dataset, the test dataset is 20% of the total dataset
X_train, X_test, y_train, y_test = train_test_split(features, target, test_size=0.2, random_state=42)

# Applying the standar scaler for the scaling unbalance scale of dataset
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

# Applying the logistic regression
linreg_model = LogisticRegression()
linreg_model.fit(X_train_scaled,y_train)

# Predicting the test dataset (scaled)
y_pred = linreg_model.predict(X_test_scaled)

# Evaluating the accuracy of the model
print(f"Accuracy: {accuracy_score(y_pred, y_test)}")

Accuracy: 0.8211382113821138

# Providing the confusion matrix

# true negatives is C_{0,0},
# false positives is C_{0,1},
# false negatives is C_{1,0},
# true positives is C_{1,1}

custom_x_labels = ['Negative', 'Positve']
custom_y_labels = ['Negative', 'Positve']

fig, ax = plt.subplots(figsize=(6,6))
sns.heatmap(confusion_matrix(y_test, y_pred),
            xticklabels=custom_x_labels,
            yticklabels=custom_y_labels,
            annot=True,
            ax=ax)
ax.set_xlabel('Predicted')
ax.set_ylabel('Actual')

Text(41.722222222222214, 0.5, 'Actual')

# Generating the classification report
print(classification_report(y_pred, y_test, target_names=['Loan Status-No', 'Loan Status-Yes']));

                 precision    recall  f1-score   support

 Loan Status-No       0.47      0.95      0.63        20
Loan Status-Yes       0.99      0.80      0.88       103

       accuracy                           0.82       123
      macro avg       0.73      0.87      0.76       123
   weighted avg       0.90      0.82      0.84       123

Akurasi model sekitar 82% untuk memprediksi target (status peminjaman). Model ini cocok untuk memprediksi target sebenarnya (status pinjaman yang dapat diterima) sensitivitas (recall) 80%, tetapi lebih sensitif untuk memprediksi target status pinjaman yang tidak dapat diterima. Namun, model ini mampu memprediksi status pinjaman yang dapat diterima dengan presisi 99%.

# Plotting the ROC curve
y_pred_proba = linreg_model.predict_proba(X_test_scaled)[::,1]
fpr, tpr, _ = roc_curve(y_test,  y_pred_proba)
auc = roc_auc_score(y_test, y_pred_proba)

fig, ax = plt.subplots(figsize=(6,6))
ax.plot(fpr,tpr,label=f"data 1, auc={auc:.2f}")
ax.set_xlabel('False Positive Rate')
ax.set_ylabel('True Positive Rate')
ax.set_title('Reciever Operating Characteristic (ROC) Curve')
plt.legend(loc=4)
plt.show()