Statistics was always a part of investment banking and stock exchange industry from last few decades. Using Statistics and Data Analytics, investment bankers, Venture Capitalists, wall-street analysts were able to identify patterns in data, do a quick analysis of company funds, and predict stock prices as well as Profits & Losses. Machine learning has been able to disrupt the BFSI industry with its ability to provide data-driven decisions.
At e-Zest, a team of Data Scientists have been working on solving problems for our clients in bank and financial institutions domain. We have been able to provide value through numerous services such as fraud detection, virtual assistants or conversational AI to reach millions of customers, customer segmentation, churn predictions to name a few.
Below are some of the examples and use cases of Machine Learning widely used in the Banking and overall BFSI domain.
JPMorgan has implemented a machine learning algorithm-based tool for understanding contract document and extracts important information from it. It was noted that this algorithm was able to gain information from 12,500 credit documents within few seconds than 360,000 person-hours!
Document information extraction is part of machine learning where OCR (Optical Character Recognition) and deep learning-based algorithms being used for extraction of very specific information from a similar type of document. At e-Zest, we have built machine learning algorithms to process financial documents such as salary slips automating the entire process.
Virtual assistants or chatbots are being used in multiple sectors. As the banking sector started providing mobile banking facilities to their customers, they have a good opportunity to use cognitive services. It was noted that Erica, the virtual assistant from Bank of America has been able to serve as an on-point financial adviser for 45 million customers across the globe.
At e-Zest, we have the capacity of building virtual assistants which can answer the queries of customers and have social conversations with them relating to the product or services they are interested in.
Fraud detection and risk management
Data analytics has been used for capturing frauds happening in the Finance industry. Though fraud cases are far less than 0.01% of whole transactions, but when they happened, they make far more losses. Other than the financial loss, they also impact sentiments of investors and markets tends to react quickly on to such incidents.
Machine learning can detect fraud before it happens. For example:
Using past history transactions of credit card, machine learning algorithm mimics the process of banking analyst to detect fraudulent transactions, create rules to detect the behavior of those customers who have a high risk of being the defaulting and provide us the likelihood of credit card being compromised. Using stronger machine learning algorithms, we can also remove false positives to make fraud detection process more efficient.
Equity & stock predictions
Wall Street investors, VC firms, Fund managers are using AI as a great tool for predicting market movements. With AI, experts can analyze public remarks, sentiment analysis then compare the same with historical data to predict stock performance in the long term and short term.
Banks consist of all segments of customers right from young, salaried, families and old age people too. Because of these varied types, their requirements, needs, attitude and behavior are unique. Banks need to segment those customers to support sales, market campaigns, and promotion efforts. Types of products for example – insurance products, investing options vary depending on their type.
Clustering techniques can segment customers based on their different behaviours, profiles and also age.
Using big data technologies and machine learning algorithms on huge datasets help in identifying patterns in data and bring important information about fraudulent transactions. Identified patterns are sent to agencies to track down relevant people.