Project 1: Predicting the probability of Loan Default in Peer to Peer lending Market - Ranking loans.
The motive of the project is to solve the problem of information asymmetry that persists in the peer to peer lending market. The objective is to help the lenders to make guided investment decision. The model ranks the loans based of probabilty of default.
Project 2: Credit card fraud detection - Comparision of various supervised learning methods and Sampling methods.
Despite technology and overall awareness, credit card fraud is on the rise and so are the different types of credit card scams. Fraud transactions contribute to heavy losses for the credit card issuers and Card holders get flagged on the credit report, may result in higher interest rates and administration fees. With the nature of the scams getting newer and more sophisticated, there is a need to better improve the performance of fraud detection models. The objective of the project is to classify a transaction as either fraudulent or legitimate based on the features of a transaction, however the challange is heavy imbalance in the data. The project explores various supervised learning methods combined with the most popular sampling methods.
Project 3: Financial time series analysis on crude oil prices (WTI) - Quantify Market Risk using VAR approaches.
Oil prices depend on many factors and have unpredictable movements. This fluctuating nature makes it harder to forecast. Volatility of oil prices has increased in the recent decade resulting in increased risk for the investors. The time variant volatility amplifies the issues when forecasting. The motivation of this project is to estimate the value at risk for the investors.
Understanding the dynamics of the oil prices is crucial to build a well fitted time series model. Furthermore, the VAR estimates using econometric and simulated approaches directly depend on the estimated model. In this paper, First, we study the dynamics of WTI crude oil in an exploratory analysis. Second, we apply various linear time series models, compare their relative performance and determine the best fitted model. Finally, measure the level of financial risk and quantify market risk using various statistical VAR approaches.
