There seems to be no end to bad loans in the country. According to the Reserve Bank of India, the overall bad loans as of March 2021 stood at INR 8.35 lakh crore, compared to INR 8.96 lakh crore in March 2020.  

In a bid to solve the loan defaulter problem, Deloitte, in partnership with MachineHack, is launching a hackathon for data scientists and machine learning practitioners called ‘Machine Learning Challenge,’ from November 29  to December 13, 2021. The winners of the hackathon will get a chance to win cash prizes worth up to INR 1 lakh. 

Let the challenge begin! 

The time is ripe to showcase your talent and technical prowess. So, what are you waiting for? Participate in the fortnight-long hackathon, and get a chance to win cash prizes, alongside improving their Global Leaderboard Rankings and becoming the ultimate MachineHack Grand Master. The challenge is open to data scientists, machine learning practitioners, analytics professionals, and tech enthusiasts. It is designed for participants who are intermediate level. 

The challenge starts on November 29, 2021. 

Problem Statement & Description 

Banks run into losses when a customer doesn’t pay their loans on time. Because of this, every year, banks have losses in crores, and this also impacts the country’s economic growth to a large extent. In this hackathon, we look at various attributes such as funded amount, location, loan, balance, etc., to predict if a person will be a loan defaulter or not. 

To solve this problem, MachineHack has created a training dataset of 67,463 rows and 35 columns and a testing dataset of 28,913 rows and 34 columns. The hackathon demands a few pre-requisite skills like big dataset, underfitting vs overfitting, and the ability to optimise “log_loss” to generalise well on unseen data. 

Datasets will be made live on November 29, 2021, at 6:00 PM.

Submission Guidelines 

• Sklearn models should support the predict() method to generate the predicted values. 
• The participant should submit a .csv file with exactly  28,913 rows with 1 column (Loan status). The submission will return an Invalid Score if you have extra rows or columns.
• The file should have exactly 1 column.

Note: Do not shuffle the sequence of the test series. 

If you are using pandas, use this submission code: 

submission_df.to_csv(‘my_submission_file.csv’, index=False)

Evaluation Criteria 

The submission will be assessed using the Log Loss metric. You can use sklearn.metric.log_loss to calculate the same.

The evaluation will be done in two stages:

Stage-1 

• It will be based on the participants standing on the private leaderboard.
• The public leaderboard uses 30% of the provided test.csv dataset to evaluate.
• The private leaderboard uses 100% of the provided test.csv dataset to evaluate. 

Stage-2 

• MCQs to test skills in SQL, data engineering, and regression, which needs to be submitted by 13th December 2021 at 6:00 PM IST.
• Not submitting the MCQ will result in disqualification.
• The final score will be a sum/aggregate of the final leaderboard, MCQ score.  

Prizes

The Machine Learning Challenge will select five (5) winners based on the given evaluation criteria. The cash prizes are as follows: 

First Prize: INR 40,000 

Second Price: INR 30,000

Third Prize: INR 10,000

Fourth Prize: INR 10,000

Fifth Prize: INR 10,000

Note: 

• Make sure your MachineHack profile is up to date with all the relevant information/details. 
• Make sure you have gone through the ‘Rules’ section before participating. 
• The participants will receive the prize money only if selected by Deloitte and MachineHack. 

The hackathon will end on December 13, 2021, at 6:00 PM.

Attribute Description:

• ID
• Loan Amount 
• Funded Amount 
• Funded Amount Investor 
• Term 
• Batch Enrolled 
• Interest Rate 
• Grade
• Sub Grade 
• Employment Duration 
• Home Ownership 
• Verification Status 
• Payment Plan
• Loan Title 
• Debit to Income 
• Delinquency – two years 
• Inquires – six months 
• Open Account 
• Public Record 
• Revolving Balance 
• Revolving Utilities 
• Total Accounts 
• Initial List Status 
• Total Received Interest 
• Total Received Late Fee 
• Recoveries 
• Collection Recovery Fee 
• Collection 12 months Medical 
• Application Type 
• Last week Pay 
• Accounts Delinquent 
• Total Collection Amount 
• Total Current Balance 
• Total Revolving Credit Limit
• Loan Status 

Skills: 

• Big dataset, underfitting vs overfitting
• Optimising log_loss to generalise well on unseen data

Winners Announcement

Final winners will be notified via email based on an aggregate score of their private leaderboard rankings.

Datasets will be made live on November 29, 2021, at 6:00 PM.

The hackathon will end on December 13, 2021, at 6:00 PM.