ieochc5ud

December 11, 2024

[390]: # This Python 3 environment comes with many helpful analytics libraries␣
↪installed

# It is defined by the kaggle/python Docker image: https://github.com/kaggle/

↪docker-python

# For example, here's several helpful packages to load

import numpy as np # linear algebra

import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)

# Input data files are available in the read-only "../input/" directory

# For example, running this (by clicking run or pressing Shift+Enter) will list␣
↪all files under the input directory

import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename))

# You can write up to 20GB to the current directory (/kaggle/working/) that␣

↪gets preserved as output when you create a version using "Save & Run All"

# You can also write temporary files to /kaggle/temp/, but they won't be saved␣
↪outside of the current session

/kaggle/input/city-u-10-f-fun-ai-final-project/10F_train.csv
/kaggle/input/city-u-10-f-fun-ai-final-project/10F_sample_submission.csv
/kaggle/input/city-u-10-f-fun-ai-final-project/10F_test.csv

[391]: import pandas as pd

# Đọc dữ liệu
train = pd.read_csv('/kaggle/input/city-u-10-f-fun-ai-final-project/10F_train.
↪csv')

test = pd.read_csv('/kaggle/input/city-u-10-f-fun-ai-final-project/10F_test.
↪csv')

[392]: train.head()

1
[392]: id CustomerId Surname CreditScore Geography Gender Age Tenure \
0 1 15749177 Okwudiliolisa 627 France Male 33.0 1
1 2 15694510 Hsueh 678 France Male 40.0 10
2 3 15741417 Kao 581 France Male 34.0 2
3 4 15766172 Chiemenam 716 Spain Male 33.0 5
4 5 15771669 Genovese 588 Germany Male 36.0 4

Balance NumOfProducts HasCrCard IsActiveMember EstimatedSalary \

0 0.00 2 1 1 49503.50
1 0.00 2 1 0 184866.69
2 148882.54 1 1 1 84560.88
3 0.00 2 1 1 15068.83
4 131778.58 1 1 0 136024.31

Exited
0 0
1 0
2 0
3 0
4 1

[393]: train.describe()

[393]: id CustomerId CreditScore Age \

count 132027.000000 1.320270e+05 132027.000000 132027.000000
mean 82432.091133 1.569183e+07 656.783832 38.120996
std 47705.125906 7.137972e+04 80.043164 8.869802
min 1.000000 1.556570e+07 350.000000 18.000000
25% 41083.500000 1.563290e+07 598.000000 32.000000
50% 82435.000000 1.569013e+07 660.000000 37.000000
75% 123790.500000 1.575662e+07 710.000000 42.000000
max 165033.000000 1.581569e+07 850.000000 92.000000

Tenure Balance NumOfProducts HasCrCard \

count 132027.000000 132027.000000 132027.000000 132027.000000
mean 5.021821 55609.625464 1.554682 0.753838
std 2.808487 62860.390849 0.547018 0.430776
min 0.000000 0.000000 1.000000 0.000000
25% 3.000000 0.000000 1.000000 1.000000
50% 5.000000 0.000000 2.000000 1.000000
75% 7.000000 120107.645000 2.000000 1.000000
max 10.000000 250898.090000 4.000000 1.000000

IsActiveMember EstimatedSalary Exited

count 132027.000000 132027.000000 132027.00000
mean 0.497300 112683.672952 0.21182
std 0.499995 50275.570007 0.40860

2
 min 0.000000 11.580000 0.00000
25% 0.000000 74835.650000 0.00000
50% 0.000000 118024.100000 0.00000
75% 1.000000 155616.750000 0.00000
max 1.000000 199992.480000 1.00000

[394]: print(train.shape)

(132027, 14)

[395]: test.head()

[395]: id CustomerId Surname CreditScore Geography Gender Age Tenure \

0 0 15674932 Okwudilichukwu 668 France Male 33 3
1 12 15717962 Rossi 759 Spain Male 71 9
2 20 15781496 Udegbulam 773 Spain Male 35 9
3 22 15759913 Trentini 553 Germany Female 43 9
4 24 15626012 Obidimkpa 714 France Male 26 6

Balance NumOfProducts HasCrCard IsActiveMember EstimatedSalary

0 0.00 2 1 0 181449.97
1 0.00 1 1 1 93081.87
2 0.00 2 0 1 87549.36
3 85200.82 1 1 0 160574.09
4 149879.66 2 1 1 50016.17

[396]: # Hiển thị thông tin tổng quan về dữ liệu

print(train.info())

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 132027 entries, 0 to 132026
Data columns (total 14 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 id 132027 non-null int64
1 CustomerId 132027 non-null int64
2 Surname 132027 non-null object
3 CreditScore 132027 non-null int64
4 Geography 132027 non-null object
5 Gender 132027 non-null object
6 Age 132027 non-null float64
7 Tenure 132027 non-null int64
8 Balance 132027 non-null float64
9 NumOfProducts 132027 non-null int64
10 HasCrCard 132027 non-null int64
11 IsActiveMember 132027 non-null int64
12 EstimatedSalary 132027 non-null float64
13 Exited 132027 non-null int64

3
 dtypes: float64(3), int64(8), object(3)
memory usage: 14.1+ MB
None

[397]: print(test.info())

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 33007 entries, 0 to 33006
Data columns (total 13 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 id 33007 non-null int64
1 CustomerId 33007 non-null int64
2 Surname 33007 non-null object
3 CreditScore 33007 non-null int64
4 Geography 33007 non-null object
5 Gender 33007 non-null object
6 Age 33007 non-null int64
7 Tenure 33007 non-null int64
8 Balance 33007 non-null float64
9 NumOfProducts 33007 non-null int64
10 HasCrCard 33007 non-null int64
11 IsActiveMember 33007 non-null int64
12 EstimatedSalary 33007 non-null float64
dtypes: float64(2), int64(8), object(3)
memory usage: 3.3+ MB
None

[398]: # Kiểm tra dữ liệu thiếu

print(train.isnull().sum())

id 0
CustomerId 0
Surname 0
CreditScore 0
Geography 0
Gender 0
Age 0
Tenure 0
Balance 0
NumOfProducts 0
HasCrCard 0
IsActiveMember 0
EstimatedSalary 0
Exited 0
dtype: int64

[399]: print(test.isnull().sum())

4
 id 0
CustomerId 0
Surname 0
CreditScore 0
Geography 0
Gender 0
Age 0
Tenure 0
Balance 0
NumOfProducts 0
HasCrCard 0
IsActiveMember 0
EstimatedSalary 0
dtype: int64

[400]: # Xem phân phối của biến Exited

print(train['Exited'].value_counts())

Exited
0 104061
1 27966
Name: count, dtype: int64

[401]: import seaborn as sns

import matplotlib.pyplot as plt

[402]: # Vẽ biểu đồ phân loại

sns.countplot(x='Exited', data=train)
plt.show()

5
[403]: cols = ['Gender','Geography','HasCrCard','IsActiveMember']

n_rows = 2
n_cols = 3

fig, ax = plt. subplots (n_rows, n_cols, figsize=(n_cols*3.5, n_rows*3.5))

for r in range(0, n_rows):

for c in range(0, n_cols):
i = r*n_cols + c #index to loop through list "cols"
if i < len(cols):
ax_i = ax[r,c]
sns. countplot (data=train, x=cols[i], hue="Exited",␣
↪palette="Blues",ax=ax_i)

ax_i.set_title(f"Figure {i+1}: Exited Rate vs {cols[i]}")

ax_i.legend(title = '', loc='upper right', labels = ['Not Exited',␣
↪'Exited'])

ax.flat[-2].set_visible(False) #Remove the last subplot

plt.tight_layout()
plt.show()

6
[404]: sns.histplot(data=train, x='Age', hue= 'Exited', bins = 40, kde=True)

/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning:
use_inf_as_na option is deprecated and will be removed in a future version.
Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)

[404]: <Axes: xlabel='Age', ylabel='Count'>

7
[405]: sns.histplot (data=train, x='CreditScore', hue = 'Exited', bins = 40)

/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning:
use_inf_as_na option is deprecated and will be removed in a future version.
Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)

8
[405]: <Axes: xlabel='CreditScore', ylabel='Count'>

[406]: sns.histplot (data=train, x='Tenure', hue = 'Exited', bins = 40)

/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning:
use_inf_as_na option is deprecated and will be removed in a future version.
Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.

9
 data_subset = grouped_data.get_group(pd_key)

[406]: <Axes: xlabel='Tenure', ylabel='Count'>

[407]: sns.histplot (data=train, x='Balance', hue = 'Exited', bins = 40)

/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning:
use_inf_as_na option is deprecated and will be removed in a future version.
Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)

[407]: <Axes: xlabel='Balance', ylabel='Count'>

10
[408]: sns.histplot (data=train, x='EstimatedSalary', hue = 'Exited', bins = 40)

/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning:
use_inf_as_na option is deprecated and will be removed in a future version.
Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)

[408]: <Axes: xlabel='EstimatedSalary', ylabel='Count'>

11
[409]: sns.histplot (data=train, x='NumOfProducts', hue = 'Exited', bins = 40)

/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning:
use_inf_as_na option is deprecated and will be removed in a future version.
Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)
/opt/conda/lib/python3.10/site-packages/seaborn/_oldcore.py:1075: FutureWarning:
When grouping with a length-1 list-like, you will need to pass a length-1 tuple
to get_group in a future version of pandas. Pass `(name,)` instead of `name` to
silence this warning.
data_subset = grouped_data.get_group(pd_key)

12
[409]: <Axes: xlabel='NumOfProducts', ylabel='Count'>

[410]: # Loại bỏ các cột không cần thiết

X = train.drop(['id', 'CustomerId', 'Surname', 'Exited'], axis=1)
y = train['Exited']
X_test = test.drop(['id', 'CustomerId', 'Surname'], axis=1)

# Xác định các cột phân loại và số

categorical_features = ['Geography','Gender','IsActiveMember']
numerical_features = ['CreditScore', 'Age','Tenure', 'Balance',
'NumOfProducts']

[411]: print (categorical_features)

print (numerical_features)

['Geography', 'Gender', 'IsActiveMember']

['CreditScore', 'Age', 'Tenure', 'Balance', 'NumOfProducts']

[412]: from sklearn.pipeline import Pipeline

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import OneHotEncoder, StandardScaler

13
 # Pipeline cho biến phân loại
categorical_transformer = Pipeline(steps=[
('onehot', OneHotEncoder(handle_unknown='ignore'))])

# Pipeline cho biến số

numerical_transformer = Pipeline(steps=[
('scaler', StandardScaler())])

# Kết hợp xử lý cả hai loại biến

preprocessor = ColumnTransformer(
transformers=[
('num', numerical_transformer, numerical_features),
('cat', categorical_transformer, categorical_features)])

[413]: # Chia tập train và validation

from sklearn.model_selection import train_test_split
X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2,␣
↪random_state=42, stratify=y)

print("Kích thước tập train:", X_train.shape)

print("Kích thước tập validation:", X_val.shape)

Kích thước tập train: (105621, 10)

Kích thước tập validation: (26406, 10)

[414]: from sklearn.neighbors import KNeighborsClassifier

from sklearn.metrics import classification_report, accuracy_score,␣
↪confusion_matrix

# Tạo pipeline cho KNN

knn_model = Pipeline(steps=[('preprocessor', preprocessor),
('classifier',␣
↪KNeighborsClassifier(n_neighbors=5))])

# Huấn luyện mô hình

knn_model.fit(X_train, y_train)

# Dự đoán trên tập validation

y_pred_knn = knn_model.predict(X_val)

# Đánh giá mô hình

print("KNN Accuracy:", accuracy_score(y_val, y_pred_knn))
print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_knn))
print("Classification Report:\n", classification_report(y_val, y_pred_knn))

KNN Accuracy: 0.8485950162841779

14
 Confusion Matrix:
[[19366 1447]
[ 2551 3042]]
Classification Report:
precision recall f1-score support

0 0.88 0.93 0.91 20813

1 0.68 0.54 0.60 5593

accuracy 0.85 26406

macro avg 0.78 0.74 0.75 26406
weighted avg 0.84 0.85 0.84 26406

[415]: #KNN test with cross validation

from sklearn.model_selection import cross_val_score
cv_scores = cross_val_score(knn_model, X_train, y_train, cv=5,␣
↪scoring='accuracy')

# In ra các kết quả từ các fold

print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")

# Huấn luyện lại mô hình trên toàn bộ dữ liệu huấn luyện và đánh giá trên tập␣
↪validation

knn_model.fit(X_train, y_train)
y_pred_knn = knn_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy on Validation Set:", accuracy_score(y_val,␣
↪y_pred_knn))

print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_knn))

print("Classification Report:\n", classification_report(y_val, y_pred_knn))

Cross-validation scores: [0.85031953 0.8500284 0.84903427 0.84605188

0.84557849]
Mean cross-validation accuracy: 0.848202515437165
Random Forest Accuracy on Validation Set: 0.8485950162841779
Confusion Matrix:
[[19366 1447]
[ 2551 3042]]
Classification Report:
precision recall f1-score support

0 0.88 0.93 0.91 20813

1 0.68 0.54 0.60 5593

15
 accuracy 0.85 26406
macro avg 0.78 0.74 0.75 26406
weighted avg 0.84 0.85 0.84 26406

[416]: from sklearn.naive_bayes import GaussianNB

# Tạo pipeline cho Naive Bayes

nb_model = Pipeline(steps=[
('preprocessor', preprocessor),
('classifier', GaussianNB())])

# Huấn luyện mô hình

nb_model.fit(X_train, y_train)

# Dự đoán trên tập validation

y_pred_nb = nb_model.predict(X_val)

# Đánh giá mô hình

print("Naive Bayes Accuracy:", accuracy_score(y_val, y_pred_nb))
print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_nb))
print("Classification Report:\n", classification_report(y_val, y_pred_nb))

Naive Bayes Accuracy: 0.80807392259335

Confusion Matrix:
[[18501 2312]
[ 2756 2837]]
Classification Report:
precision recall f1-score support

0 0.87 0.89 0.88 20813

1 0.55 0.51 0.53 5593

accuracy 0.81 26406

macro avg 0.71 0.70 0.70 26406
weighted avg 0.80 0.81 0.81 26406

[417]: #Naive Bayes test with cross validation

from sklearn.model_selection import cross_val_score
cv_scores = cross_val_score(nb_model, X_train, y_train, cv=5,␣
↪scoring='accuracy')

# In ra các kết quả từ các fold

print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")

16
 # Huấn luyện lại mô hình trên toàn bộ dữ liệu huấn luyện và đánh giá trên tập␣
↪validation

nb_model.fit(X_train, y_train)
y_pred_nb = nb_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy on Validation Set:", accuracy_score(y_val,␣
↪y_pred_nb))

print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_nb))

print("Classification Report:\n", classification_report(y_val, y_pred_nb))

Cross-validation scores: [0.80426036 0.80444045 0.80818027 0.80604999

0.81035789]
Mean cross-validation accuracy: 0.8066577896198159
Random Forest Accuracy on Validation Set: 0.80807392259335
Confusion Matrix:
[[18501 2312]
[ 2756 2837]]
Classification Report:
precision recall f1-score support

0 0.87 0.89 0.88 20813

1 0.55 0.51 0.53 5593

accuracy 0.81 26406

macro avg 0.71 0.70 0.70 26406
weighted avg 0.80 0.81 0.81 26406

[418]: from sklearn.linear_model import LogisticRegression

# Tạo pipeline cho Logistic Regression

lr_model = Pipeline(steps=[
('preprocessor', preprocessor),
('classifier', LogisticRegression(random_state=42, max_iter=1000))])

# Huấn luyện mô hình

lr_model.fit(X_train, y_train)

# Dự đoán trên tập validation

y_pred_lr = lr_model.predict(X_val)

# Đánh giá mô hình

print("Logistic Regression Accuracy:", accuracy_score(y_val, y_pred_lr))
print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_lr))
print("Classification Report:\n", classification_report(y_val, y_pred_lr))

17
 Logistic Regression Accuracy: 0.8328410209800803
Confusion Matrix:
[[19887 926]
[ 3488 2105]]
Classification Report:
precision recall f1-score support

0 0.85 0.96 0.90 20813

1 0.69 0.38 0.49 5593

accuracy 0.83 26406

macro avg 0.77 0.67 0.69 26406
weighted avg 0.82 0.83 0.81 26406

[419]: #Linear test with cross validation

from sklearn.model_selection import cross_val_score
cv_scores = cross_val_score(lr_model, X_train, y_train, cv=5,␣
↪scoring='accuracy')

# In ra các kết quả từ các fold

print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")

# Huấn luyện lại mô hình trên toàn bộ dữ liệu huấn luyện và đánh giá trên tập␣
↪validation

lr_model.fit(X_train, y_train)
y_pred_lr = lr_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy on Validation Set:", accuracy_score(y_val,␣
↪y_pred_lr))

print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_lr))

print("Classification Report:\n", classification_report(y_val, y_pred_lr))

Cross-validation scores: [0.83427219 0.83132929 0.83274948 0.834217

0.83724673]
Mean cross-validation accuracy: 0.8339629400474402
Random Forest Accuracy on Validation Set: 0.8328410209800803
Confusion Matrix:
[[19887 926]
[ 3488 2105]]
Classification Report:
precision recall f1-score support

0 0.85 0.96 0.90 20813

1 0.69 0.38 0.49 5593

18
 accuracy 0.83 26406
macro avg 0.77 0.67 0.69 26406
weighted avg 0.82 0.83 0.81 26406

[420]: from sklearn.tree import DecisionTreeClassifier

# Tạo pipeline cho Decision Tree

dt_model = Pipeline(steps=[
('preprocessor', preprocessor),
('classifier', DecisionTreeClassifier(random_state=42, max_depth=10))])

# Huấn luyện mô hình

dt_model.fit(X_train, y_train)

# Dự đoán trên tập validation

y_pred_dt = dt_model.predict(X_val)

# Đánh giá mô hình

print("Decision Tree Accuracy:", accuracy_score(y_val, y_pred_dt))
print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_dt))
print("Classification Report:\n", classification_report(y_val, y_pred_dt))

Decision Tree Accuracy: 0.860978565477543

Confusion Matrix:
[[19580 1233]
[ 2438 3155]]
Classification Report:
precision recall f1-score support

0 0.89 0.94 0.91 20813

1 0.72 0.56 0.63 5593

accuracy 0.86 26406

macro avg 0.80 0.75 0.77 26406
weighted avg 0.85 0.86 0.85 26406

[421]: #Decision tree test with cross validation

from sklearn.model_selection import cross_val_score
cv_scores = cross_val_score(dt_model, X_train, y_train, cv=5,␣
↪scoring='accuracy')

# In ra các kết quả từ các fold

print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")

19
 # Huấn luyện lại mô hình trên toàn bộ dữ liệu huấn luyện và đánh giá trên tập␣
↪validation

dt_model.fit(X_train, y_train)
y_pred_dt = dt_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy on Validation Set:", accuracy_score(y_val,␣
↪y_pred_dt))

print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_dt))

print("Classification Report:\n", classification_report(y_val, y_pred_dt))

Cross-validation scores: [0.85921893 0.85892823 0.85646658 0.85826548

0.85599318]
Mean cross-validation accuracy: 0.8577744819263879
Random Forest Accuracy on Validation Set: 0.860978565477543
Confusion Matrix:
[[19580 1233]
[ 2438 3155]]
Classification Report:
precision recall f1-score support

0 0.89 0.94 0.91 20813

1 0.72 0.56 0.63 5593

accuracy 0.86 26406

macro avg 0.80 0.75 0.77 26406
weighted avg 0.85 0.86 0.85 26406

[422]: #Randomforest test

from sklearn.ensemble import RandomForestClassifier

# Tạo pipeline cho Random Forest

rf_model = Pipeline(steps=[
('preprocessor', preprocessor),
('classifier', RandomForestClassifier(random_state=42, n_estimators=100,␣
↪max_depth=10))])

# Huấn luyện mô hình

rf_model.fit(X_train, y_train)

# Dự đoán trên tập validation

y_pred_rf = rf_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy:", accuracy_score(y_val, y_pred_rf))

20
 print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_rf))
print("Classification Report:\n", classification_report(y_val, y_pred_rf))

Random Forest Accuracy: 0.8629856850715747

Confusion Matrix:
[[19879 934]
[ 2684 2909]]
Classification Report:
precision recall f1-score support

0 0.88 0.96 0.92 20813

1 0.76 0.52 0.62 5593

accuracy 0.86 26406

macro avg 0.82 0.74 0.77 26406
weighted avg 0.85 0.86 0.85 26406

[423]: #RandomForesr test with cross validation

from sklearn.model_selection import cross_val_score
cv_scores = cross_val_score(rf_model, X_train, y_train, cv=5,␣
↪scoring='accuracy')

# In ra các kết quả từ các fold

print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")

# Huấn luyện lại mô hình trên toàn bộ dữ liệu huấn luyện và đánh giá trên tập␣
↪validation

rf_model.fit(X_train, y_train)
y_pred_nb = rf_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy on Validation Set:", accuracy_score(y_val,␣
↪y_pred_rf))

print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_rf))

print("Classification Report:\n", classification_report(y_val, y_pred_rf))

Cross-validation scores: [0.86338462 0.86295209 0.86209998 0.86219466

0.86318879]
Mean cross-validation accuracy: 0.8627640277919392
Random Forest Accuracy on Validation Set: 0.8629856850715747
Confusion Matrix:
[[19879 934]
[ 2684 2909]]
Classification Report:
precision recall f1-score support

21
 0 0.88 0.96 0.92 20813
1 0.76 0.52 0.62 5593

accuracy 0.86 26406

macro avg 0.82 0.74 0.77 26406
weighted avg 0.85 0.86 0.85 26406

[424]: from xgboost import XGBClassifier

from sklearn.pipeline import Pipeline
from sklearn.metrics import accuracy_score, confusion_matrix,␣
↪classification_report

# Tạo pipeline cho XGBoost

xgb_model = Pipeline(steps=[
('preprocessor', preprocessor), # Preprocessing (nếu có)
('classifier', XGBClassifier(random_state=42, use_label_encoder=False,␣
↪eval_metric='mlogloss'))]) # Khởi tạo mô hình XGBoost

# Huấn luyện mô hình

xgb_model.fit(X_train, y_train)

# Dự đoán trên tập validation

y_pred_xgb = xgb_model.predict(X_val)

# Đánh giá mô hình

print("XGBoost Accuracy:", accuracy_score(y_val, y_pred_xgb))
print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_xgb))
print("Classification Report:\n", classification_report(y_val, y_pred_xgb))

XGBoost Accuracy: 0.865295766113762

Confusion Matrix:
[[19702 1111]
[ 2446 3147]]
Classification Report:
precision recall f1-score support

0 0.89 0.95 0.92 20813

1 0.74 0.56 0.64 5593

accuracy 0.87 26406

macro avg 0.81 0.75 0.78 26406
weighted avg 0.86 0.87 0.86 26406

22
[425]: #XGBClassifier test with cross validation
from sklearn.model_selection import cross_val_score
cv_scores = cross_val_score(nb_model, X_train, y_train, cv=5,␣
↪scoring='accuracy')

# In ra các kết quả từ các fold

print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")

# Huấn luyện lại mô hình trên toàn bộ dữ liệu huấn luyện và đánh giá trên tập␣
↪validation

xgb_model.fit(X_train, y_train)
y_pred_nb = xgb_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy on Validation Set:", accuracy_score(y_val,␣
↪y_pred_xgb))

print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_xgb))

print("Classification Report:\n", classification_report(y_val, y_pred_xgb))

Cross-validation scores: [0.80426036 0.80444045 0.80818027 0.80604999

0.81035789]
Mean cross-validation accuracy: 0.8066577896198159
Random Forest Accuracy on Validation Set: 0.865295766113762
Confusion Matrix:
[[19702 1111]
[ 2446 3147]]
Classification Report:
precision recall f1-score support

0 0.89 0.95 0.92 20813

1 0.74 0.56 0.64 5593

accuracy 0.87 26406

macro avg 0.81 0.75 0.78 26406
weighted avg 0.86 0.87 0.86 26406

[ ]: pip install lightgbm

Requirement already satisfied: lightgbm in /opt/conda/lib/python3.10/site-

packages (4.2.0)
Requirement already satisfied: numpy in /opt/conda/lib/python3.10/site-packages
(from lightgbm) (1.26.4)
Requirement already satisfied: scipy in /opt/conda/lib/python3.10/site-packages
(from lightgbm) (1.14.1)

23
[ ]: #LGBMClassifier test
from lightgbm import LGBMClassifier
from sklearn.pipeline import Pipeline
from sklearn.metrics import accuracy_score, confusion_matrix,␣
↪classification_report

# Tạo pipeline cho LightGBM

lgbm_model = Pipeline(steps=[
('preprocessor', preprocessor), # Preprocessing (nếu có)
('classifier', LGBMClassifier(random_state=42))]) # Khởi tạo mô hình␣
↪LightGBM

# Huấn luyện mô hình

lgbm_model.fit(X_train, y_train)

# Dự đoán trên tập validation

y_pred_lgbm = lgbm_model.predict(X_val)

# Đánh giá mô hình

print("LightGBM Accuracy:", accuracy_score(y_val, y_pred_lgbm))
print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_lgbm))
print("Classification Report:\n", classification_report(y_val, y_pred_lgbm))

[ ]: #LightGBM test with cross validation

from sklearn.model_selection import cross_val_score
cv_scores = cross_val_score(nb_model, X_train, y_train, cv=5,␣
↪scoring='accuracy')

# In ra các kết quả từ các fold

print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")

# Huấn luyện lại mô hình trên toàn bộ dữ liệu huấn luyện và đánh giá trên tập␣
↪validation

lgbm_model.fit(X_train, y_train)
y_pred_lgbm = lgbm_model.predict(X_val)

# Đánh giá mô hình

print("Random Forest Accuracy on Validation Set:", accuracy_score(y_val,␣
↪y_pred_lgbm))

print("Confusion Matrix:\n", confusion_matrix(y_val, y_pred_lgbm))

print("Classification Report:\n", classification_report(y_val, y_pred_lgbm))

[ ]: models = {
"KNN": y_pred_knn,
"Naive Bayes": y_pred_nb,
"Logistic Regression": y_pred_lr,

24
 "Decision Tree": y_pred_dt,
"Random Forest": y_pred_rf,
"XGBClassifier": y_pred_xgb,
"LGBMClassifier": y_pred_lgbm
}

for name, preds in models.items():

print(f"{name} Model")
print("Accuracy:", accuracy_score(y_val, preds))
print(f"Cross-validation scores: {cv_scores}")
print(f"Mean cross-validation accuracy: {cv_scores.mean()}")
print("-" * 40)

[ ]: # Dự đoán trên tập test

y_pred_test = lgbm_model.predict(X_test)

# Tạo DataFrame cho kết quả dự đoán

submission = pd.DataFrame({
'id': test['id'], # Lấy CustomerId từ dữ liệu test, kiểm tra lại tên cột␣
↪'id'

'Exited': y_pred_test # Lấy kết quả dự đoán

})

# Đảm bảo rằng tên cột chính xác

submission.columns = ['id', 'Exited']

# Lưu kết quả vào file CSV (đảm bảo lưu đúng đường dẫn trong môi trường Kaggle)
submission.to_csv('submission.csv', index=False)

# Xem trước file submission

submission.head()

[ ]: # Xem trước nội dung file submission

submission.head()

25