Scribd
Upload
85 views10 pages

Architecture Credit Card Default Prediction

The document describes the architecture for a credit card default prediction system. It details the data collection from a Kaggle dataset, exploratory data analysis, feature engineering, model creation using algorithms like logistic regression and random forests, model performance evaluation, and model deployment to a cloud platform with a user interface.

Uploaded by

Crack You
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
85 views10 pages

Architecture Credit Card Default Prediction

The document describes the architecture for a credit card default prediction system. It details the data collection from a Kaggle dataset, exploratory data analysis, feature engineering, model creation using algorithms like logistic regression and random forests, model performance evaluation, and model deployment to a cloud platform with a user interface.

Uploaded by

Crack You
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

Architecture

Credit Card Default Prediction

Architecture

Written By Bhavika Pathak


Document Version 1.0
Last Revised Date 2-Feb-2024

Credit Card Default Prediction Page 1


Architecture

Document Control
Change Records:
Version Date Author Comments
1.0 22-Feb-24 Bhavika Pathak Introduction &
Architecture
defined

Reviews:
Version Date Reviewer Comments

Approval Status:
Version Review Date Reviewed By Approved Comments
By

Credit Card Default Prediction Page 2


Architecture

Contents
1. Introduction

1.1. What is Low-Level design document?

1.2. Scope

2. Architecture

3. Architecture Description

3.1.Database (Kaggle)

3.2. Data Collection

3.3. Exploratory Data Analysis (EDA)

3.4. Feature Engineering

3.5. Feature Selection

3.6. Model Creation

3.7. Model Performance

3.8. Model Saving

3.9. User Interface Designing

3.10. Flask API Development

3.11. User Data Validation

3.12. Cloud Setup

3.13. Model Deployment

4. Unit Test Cases

Credit Card Default Prediction Page 3


Architecture

1. Introduction
1.1. What is Low-Level design document?
The goal of LLD or a low-level design document (LLD) is to give the internal logical
design of the actual program code for Credit Card Default Prediction System. LLD
describe the class diagrams with the methods and relations between classes and
program specs. It describes the modules so that the programmer can directly
code the program from the document.

1.2. Scope
Low-level design (LLD) is a component-level design process that follows a step-
by-step refinement process. This process can be used for designing data
structures, required software architecture, source code and ultimately,
performance algorithms. Overall, the data organization may be defined during
requirement analysis and then refined during data design work.

Credit Card Default Prediction Page 4


Architecture

2. Architecture

Credit Card Default Prediction Page 5


Architecture

3. Architecture Description
3.1 Database (Kaggle)
Credit Card Default Prediction Page 6
Architecture

This dataset contains information on default payments, demographic factors,


credit data, history of payment, and bill statements of credit card clients in
Taiwan from April 2005 to September 2005.

Below is the link to access the datasets present in Kaggle :

https://www.kaggle.com/datasets/uciml/default-of-credit-card-clients-
dataset/data

Columns present in datasets are:

ID,LIMIT_BAL,SEX,EDUCATION,MARRIAGE,AGE,PAY_0,PAY_2,PAY_3,PAY_4,PAY_5,
PAY_6,BILL_AMT1,BILL_AMT2,BILL_AMT3,BILL_AMT4,BILL_AMT5,BILL_AMT6,PAY
_AMT1,PAY_AMT2,PAY_AMT3,PAY_AMT4,PAY_AMT5,PAY_AMT6,default.paymen
t.next.month.

3.2 Data Collection


Obtain relevant datasets containing historical credit card transaction data,
including information on customers, their transactions, payment history, credit
limits, etc.

3.3 Exploratory Data Analysis (EDA)


In the Cleaning process, we cleaned up all the data which is not present in format
which can be recognized by machine learning algos. Also, dropped the features
which are not relevant. We handled the Null values and replaced with median
value. We visualized the data using different libraries like seaborn, matplotlib, etc.

3.4 Feature Engineering


We replaced the outlier with median which was present in the age feature in the
dataset. Also performed One hot and label encoding to convert the categorical
features to numerical Features. We transformed the data using different
transformations like ‘log’ and ‘square root’ transformations to remove the
skewness of the data present in different features.

Credit Card Default Prediction Page 7


Architecture

3.5 Feature Selection


We dropped the columns which were not required after transforming the data.
We also checked the multicollinearity between the independent features using
correlation and VIF techniques. Highly correlated features were removed after
this process.

3.6 Model Creation


After completing the feature selection process, data was split between train and
test data. Dataset was imbalanced having high no. of negative classes in
comparison to other classes therefore used SMOTE technique to handle it.
Different machine learning algorithms like Logistic Regression, DecisionTree, SVM,
RandomForest, K-NN were applied on the training data and testing data.

3.7 Model Performance


After fitting the data on different models, we checked the performance using
different metrics like accuracy score, precision, recall. We got accuracy 86.65 %.

3.8 Model Saving


After comparing all accuracies, we have chosen RandomForest as our best model
and so we have dumped this model in a pickle file format.

3.9 User Interface Designing


We created a user interactive page where user can enter their input values to
different fields of our application. In these front-end page, we designed a form
which is taking values using text-boxes. These html user input data is transferred
in json format to backend. After filling the fields values, user can see the result by
clicking on the predict button.

3.10 Test Case Validation


Data validation is done here which was entered by user.

Credit Card Default Prediction Page 8


Architecture

3.11 Cloud Setup


We are using netlify cloud to deploy our application.

3.12 Model Deployment


We deployed our model to Netlify which generate one URL. Using that URL, user
can access the User Interface of the application.

4. Unit Test Cases


Credit Card Default Prediction Page 9
Architecture

Test Case Pre-requisite Expected Result


Description
Verify whether the Application URL should Application URL should
Application URL is be defined. be accessible to the user
accessible to the user.

Verify whether the Application URL is The Application should


Application loads accessible. load completely for the
completely Application is deployed. user when the URL is
for the user when the accessed.
URL is
accessed

Verify whether user is Handled test cases at User should be able to


giving backends. see successfully valid
standard input. results.

Verify whether user is Application is accessible. User should be able to


able to edit all
edit all input fields input fields.

Verify whether user gets Application is accessible. User should get Submit
Predict button to submit button to submit the
the inputs.
inputs

Verify Predicted result is Application is accessible. User should be able to


displayed see the
predicted result

Credit Card Default Prediction Page 10

You might also like