Consists Banking Financial Institutes and Non-Banking Financial Institutes related.

• An Instutions are those corporate leaders who accepts deposits in all form, leand loans in all form, provides investment facilities also those called Banking Finacial Institutions.
• An Institutions are those corporate houses which allows to lend loans in all form (almost) and also provides investment facilities in different stocks & bonds, mutual funds, these institutions called as Non-Banking Financial Institutions.

• Commercial Banks

  • accepts deposits, offers checking account services, makes business, personal, and mortgage loans, and offers basic financial products like certificates of deposit (CDs) and savings accounts to individuals and small businesses.

• Investment Banks

  • Specialized in providing services designed to facilitate business operations such as capital expenditure financing and equity offerings, which includes IPOs [initial public offerings].
  • Also offer brokerage services for investors for trading exchanges and manage mergers & acquisitions, and other corporate restructurings.

1. Top-Up Loan Up-Sell Prediction - An Classification > link - https://github.com/shribiyani/Top-up-Loan-Prediction - Here Company wants to issue Top-Up Loan Facilities to their existing customers - It's a Multi-Class Classifier Problem. - I have two datasets - 1. Customer's Demographic Data with bank and 2. Customer's Bureau Data regarding

2. Credit Card Default Prediction - link https://github.com/shribiyani/Credit-Card-Default-Prediction

   • Here, Company wants to Predict the Credit CardHolder will be Default in next month or not on the basis of BILL_AMT and PAY_AMT history for last 6 month
   • Dataset has 30000 instances and 23 independent attributes.
   • Response class is Imbalance Class with Class 0 = 23362 and Class 1 = 6636
   • I used Class Probability to predict cardholder will be defaulter or not.
   • I have used Decision Tree, Random Forest, Extra Tree, AdaBoost, GradientBoosting and Light GBM classifiers.
   • LightGBM gave me approx. 80% ROC-AUC score chances of class 1
   • while my Accuracy is approximatly 77% and F2 Score is upto 60%.

3. Loan Approval Prediction - link https://github.com/shribiyani/Loan_Approval-Prediction/blob/main/Loan_Prediction.ipynb

   • Here, company wants to Predict the Loan Approval
   • Dataset is small but very informatic.
   • it consists 12 independent features and 1 response variable
   • Target Variable is distributed into 2:1 means We have approved class [Y] more then **rejected class [N].
   • Accurancy was upto 80%

4. Home Credit Default Risk Prediction - An Classification - Work-in Process

   • Here, Dataset is taken from Kaggle.com
   • Task - Predict Home Credit Deafult Risk
   • Data consists multiple datasets like loan balance, credit card balnce, installment details, buerau information, etc., around 7 datasets
   • Main Data(train) consists around 121 independent variables and 1 response variable.
   • Data is Imbalanced in nature.
   • Project planning as - Set-up Architecture, EDA, Extract extra features from other datasets and run complete ML pipeline.

5. Credit Card Fraud Detection - Applied Cluster Based Algorithms (Problem Based on an Classification) - work-in process

   • This data is also from Kaggle.com
   • Task - Detect Fraud and create an system that can mitigate it at ral time.
   • Data is Imbalanced in nature.
   • Project planning as - Set-up Architecture, EDA, Extract extra features from other datasets and run complete ML pipeline.