Credit Card Fraud Detection

For this case study, we have taken data of transactions made by credit cards in September 2013 by European Cardholders from Kaggle.com: Credit Card Fraud Detection Data

The datasets contains transactions made by credit cards in September 2013 by European cardholders. This dataset contains transactions that occurred in two days, where we have 492 frauds out of 284,807 transactions. The dataset is highly unbalanced, the positive class (frauds) account for 0.172% of all transactions.

The data set comprises 31 columns. All columns are anonymized except 'Time', 'Amount' and 'Feature'.

The result is "Feature" (last column) representing whether the transaction is fraudulent or not. Feature 1 means Fraud and 0 means 'No Fraud'

Can we detect any type of fraud?
Yes, we can detect any type of fraud provided the required data are available. However, please note that we may have to feature engineering to improve the overall fraud detection rate.

For further details, please click on the link  Continue to Read

Risky Loan Detection

As we could not get public dataset for Risky Loan, we have created synthetic data of 1,00,000 personal loan application dataset with columns Applicant ID, Loan Amount, Tenure, Birth Date, Age in Years, Net Take Home Pay, Home EMI, Other EMI, Credit Card Due, Total EMI and Due (Amount) and Net Salary

The data set contains 1,00,000 rows (applicants) and 11 columns.

Can we detect risky loan applicants from any type of loan data??
Yes, we can detect risky loan applicants from any type of loan provided the required data are available. However, please note that we may have to feature engineering to improve the overall risky loan detection rate.

For further details, please click on the link  Continue to Read

NPA Early Detection

Non-Performing Assets (NPA) is most serious problem faced by the Indian Banking Sector. As of March 31, 2018, provisional estimates suggest that the total volume of gross NPAs in the economy stands at Rs 10.35 lakh crore. About 85% of these NPAs are from loans and advances of public sector banks. In this scenario, there is urgent need of a solution that can detect the NPA at early stage, so that banks can take effective steps to curb it.

As we could not get public dataset for NPA, we have created synthetic data of 1,000 cash credit accounts with the following columns:
(1) AccountNo
(2) CrSummationLastMonth: Credit Summation of Last Month
(3) CrSummationCurMonth: Credit Summation of Current Month
(4) SummationPer: Percent of Credit Summation using formula: Column3 * 100 / Column2
(5) AvgBalance: Average Balance in the Current Month (All the amount are debit amount. Account with Credit Balance are excluded)
(6) CreditLimit: Credit Limit of Account Holder
(7) PercentUtilisation: Percent Utilization of Credit Limit computed using formula: Column5 * 100 / Column6
(8) CreditLimitExceededNo: Number of times credit limit exceeded in the Current Month
(9) ChequeDepositReturnedNo: Number of Cheques Returned (Deposited by the Account Holder). The account holder is payee.
(10) ChequeDepositReturnedAmount: Total Amount of Cheques Returned (Deposited by the Account Holder)
(11) ChequeReturnedNo: Number of Cheques Returned (Presented for debit to the Account). The account holder is drawer. This includes online debit received in the account and transaction was not successful because of insufficient balance or some other rasons.
(12) ChequeReturnedAmount: Amount of Cheques Returned (Presented for debit to the Account).

The data set contains 1,000 rows (Number of Cash Credit Accounts in a Branch) and 12 columns.

Can "Discover" detect likely NPA at early from any type of loan data??
Yes, "Discover" can detect likely NPA from any type of loan/account provided the required data are available. However, please note that we may have to feature engineering to improve the overall risky loan detection rate.

For further details, please click on the link  Continue to Read