acm-vit / good-client-bad-client

Help us build a Credit Card Approval system - using Machine Learning!

Kicking Off Hacktoberfest with ACM-VIT!

Good Client, Bad Client

Help us build a Credit Card Approval System - Using Machine Learning!

Overview

The main motive of the project is to build a machine learning model to predict if an applicant is 'good' or 'bad' client, different from other tasks, the definition of 'good' or 'bad' is not given.

Credit score cards are a common risk control method in the financial industry. It uses personal information and data submitted by credit card applicants to predict the probability of future defaults and credit card borrowings. The bank is able to decide whether to issue a credit card to the applicant. Credit scores can objectively quantify the magnitude of risk.

In dataset,application_record.csv is the table/file that has information about all the customers regarding their socio-economic status and credit_record.csv is the file/table that has all the payment/default records for a given client.

---

Usage

Run the following command to install all the required packages for this project

pip install -r requirements.txt

Lets get started!

 git remote add
 git fetch 
 git merge

Dataset

Link to the data set is here.

---

Submitting a Pull Request

• Fork the repository by clicking the fork button on top right corner of the page
• Clone the target repository. To clone, click on the clone button and copy the https address. Then run

git clone [HTTPS-ADDRESS]

• Go to the cloned directory by running

cd [NAME-OF-REPO]

• Create a new branch. Use

 git checkout -b [YOUR-BRANCH-NAME]

• Make your changes to the code. Add changes to your branch by using

git add .

• Commit the chanes by executing

git commit -m "your msg"

• Push to remote. To do this, run

git push origin [YOUR-BRANCH-NAME]

• Create a pull request. Go to the target repository and click on the "Compare & pull request" button. Make sure your PR description mentions which issues you're solving.
• Wait for your request to be accepted.

---

Guidelines for Pull Request

• Avoid pull requests that :
  • are automated or scripted
  • that are plagarized from someone else's branch
• Do not spam
• Project maintainer's decision on validity of PR is final.

For additional guidelines, refer to participation rules

---

What counts as a PR?

Check out our issues and try to solve them !

---

Interacting with Issues

• There are helper issues that detail all you have to do to complete the project.
  • Read the helper issues and work on the corresponding code in your fork of the repo.
  • If you have some doubt regarding the 'help' given, comment below the issue.
  • If you have some doubt not related to any 'helper issue/s' open, Open up a new issue, select doubt and fill in the template.
• If you want to provide some extra help to fellow participants, open up a new helper issue. Don't include any solution/code!
• Do not spam

---

Authors

Authors: Aryan Vats, Aditya Nalini, Varun Srinivasan

good-client-bad-client's Issues

STEP 4: Handling Missing Values

Task
Check to see if there are any missing values in both the datasets imported.
If yes, then fill those missing values.

Functions to implement
missing_values_table(df)
solution_missing_values(df)

STEP 3: Feature Creation

Task
Define the feature variables which are to be predicted using this model

Function to implement
feature_creation()

STEP 9: Evaluate model

Task
Compare the predicted and observed label classes to see the actual accuracy.

STEP 1: Import and read app data

Let's start on this project!

To create your first PR:

• Fork this repo
• Create a file with your name (yourname.py) in the folder submissions (Create one if folder is not there)
• Import the dataset and define the read_app_data() to read application data
• Make a pull request referencing this issue!

Function to Implement
read_app_data()

Setup Template Py Script

• make sure there is a template code set up
• for issues, you could create empty functions (for ex: def split(df) could return dataset split into test and train, and you could leave function description empty for collaborators to solve)

Step 5: EDA and Vintage Analysis

EDA and Vintage Analysis
Perform EDA for the data set to find best factors to be considered for the model.
What is Vintage Analysis could be searched here.

Where to show
Make all the Analysis under the Observation Heading.

STEP 2: Import the credit card dataset

Got your first PR merged? Awesome!
Continuing the task we started in our last issue:

Task
Try importing the Credit Card data set using the pandas package

STEP 10: Optimize your model

Congratulations! By now, you have successfully created a model and evaluated it, but is it the end? Of course not!
Let's optimize our model :)

Task

• Tune model parameters
  Considering, we have a relatively small size of the data and features, set high number of parameters for tuning.

• Optimize model classifier
  Fit the model with the tuned parameters and see the improvement in the accuracy of the model.

• Evaluate optimized model on testing sample
  Predict using the new-found accuracy!

STEP 6: Get started with model creation!

Completed and merged your last 5 tasks? Great! By now, you must have learnt a lot but the real 'Machine Learning' begins from here. From now on, you will be provided with the topics you need to implement.

Task
Import packages for the model and prepare data
(Check the template code file for the topics above)

STEP 7: Split and test sample data

Task
Provide code to split sample data provided into two sets, training data and testing data.

STEP 8: Define, Fit and Predict!

Task

• Define the ml model that you will be using in this project.
• Run the model on training data (Fitting)
• Predict outcomes using fitted model

Update README.md

• update description section
• upload dataset link
• mention steps to run code (under Usage)