Data in its raw form is nothing but a bundle of 0s and 1s for a machine unless the right applications are applied to analyse this data in order to enhance human expertise further. That’s what machine learning algorithms are all about. And the financial services sector is now experimenting this concept at a deeper level than ever.
“Machine learning technology holds the potential to cause the next evolution. As the costs of ML hardware falls, even smaller financial services companies are making a beeline to adopt such solutions,” Prabhakar Tiwari, chief marketing officer, Angel Broking told Inc42.
Going by the facts, according to the Research On Global Markets’ analysts, the global machine learning market will expand at a compounded annual growth rate (CAGR) of 48.3% during the 2018-2023 period to reach approximately $19.40 Bn by 2023.
In India as well, analysts believe that ML is an emerging technology under the branch of artificial intelligence (AI) and is being adopted aggressively by retail, transportation (especially airlines), and financial services companies operating in India. In 2018, this resulted in the creation of ~0.18 Mn to ~0.2 Mn new jobs, for professionals who have the right skills and expertise in ML applications.
What Is A Machine Learning Algorithm?
There are several definitions of machine learning. Here’s one for reference. A machine learning algorithm, also known as a model, is a mathematical expression that represents data in the context of a problem, often a business problem. The aim is to go from data to insight.
In simple terms, machine learning is a concept that allows a machine to learn from the data fed into it, without being explicitly programmed for it. It learns to take decisions based on the datasets it is trained on.
What Is The Difference Between AI And Machine Learning Models?
As Bernard Marr said in one of his recent posts on Forbes, artificial intelligence is the broader concept of machines being able to carry out tasks in a way that humans would consider “smart”. On the other hand, machine learning is a current application of AI-based around the idea that we should really just be able to give machines access to data and let them learn for themselves.
In short, while AI is more about programming and getting the instructions followed, machine learning encourages self-learning by machines basis the data feed given and experiences over time.
How Does Machine Learning Work?
Machine learning algorithms are trained using a training dataset to create a model. When new input data is introduced to the ML algorithm, it makes a prediction on the basis of the model that it has been trained on.
The prediction is evaluated for accuracy and if the accuracy is acceptable, the algorithm is deployed. If the accuracy is not acceptable, the ML algorithm is trained again and again with an augmented training data set.
Image Source: Edureka
Machine learning concept came into light when Arthur Samuel in 1959 suggested that rather than teaching computers everything they need to know about the world and how to carry out tasks, it might be possible to teach them to learn for themselves.
“Later, the emergence of internet and concepts like big data analytics leading to a large amount of digital information being stored and analysed played a key role in the rise of machine learning algorithms,” Marr had added.
Then came the neural networks — a computer system designed to work by classifying information in the same way a human brain does and natural language processing (NLP) — another field of AI that relies heavily on machine learning.
What Are The Use Cases Of Machine Learning In Finance?
Here are a few use cases where machine learning algorithms are being used in the finance sector:
- Fraud detection and prevention
- Customer service and recommendation
- Cluster-based target approaching
- Stock market predictions
- Algorithmic trading
- Network security
- Chatbots and virtual assistants
- Identify new data sources and leverage dark data
- Risk management
How Are Fintech Startups Using Machine Learning?
Here are a few startups which are using these machine learning use cases in streamlining day to day processes:
InCred has developed an algorithm-driven and machine learning-enabled risk management engine which enables it to build bespoke risk scorecards based on alternative data sources targeting the new-to-credit category of customers. The company believes in bringing a human approach to lending and is focussed on taking a new borrower focussed approach to the entire lending process.
FinBox is a fintech startup has built a credit risk management platform with proprietary data connectors to supercharge lending to self-employed merchants and new to credit customers. Its data connector APIs – DeviceConnect & BankConnect — bring in alternative sources of data to the platform enabling automated credit decisions. Founded by Rajat Deshpande, Anant Deshpande, Nikhil Bhawsinka and Srijan Nagar, FinBox claims to be able to leverage their dark data and consume new data sources to significantly improve credit risk performance while enhancing user adoption.
Datacultr (a PaaS for consumer lending companies) aims to help reduce the risk of fraud in consumer lending through its predictive fraud management solution. With the help of machine learning, the platform can identify potential fraudulent transactions by building models around usage, behavioural and geospatial patterns, and take immediate action.
For instance, Datacultr has turned a smartphone into a tool for financial inclusion by effectively making it into a virtual collateral. With the proprietary technology provided by Datacultr, consumer lending companies can set up triggers that can alert them against potential frauds and asset resale. It shall help in making the collection process seamless by monitoring such triggers on a smartphone.
Jonathan Bill, cofounder and CEO of CreditMate, told Inc42 in a recent interaction that fraud detection, propensity models, risk analytics and chatbot assistants are necessities of lending in this new era. As the firm deals with a rigorous process of collection which is intensely a decision-driven process and involves human intervention, Creditmate is using AI and ML to extract all of that individual knowledge and make a single ultra-smart and evolving intelligence layer. Its recommender systems match content to a user, while behavioural models predict strategy for collection resolutions and propensity models drive efficiency.
“By doing this we are increasing the amount of resolution of bad debt through automation, significantly improving efficiency in last mile or filed collections and thus making lenders of all sizes more able to lend and fuel India’s economy,” added Bill.
While Angel Broking’s Tiwari didn’t tell exactly how the company is using machine learning algorithms, he did emphasise on the fact that by identifying a target variable and using historical data to train a machine learning model, we can predict the variable’s value as close as possible to its actual value over different time periods and market conditions
“ML-based trading systems have the ability to completely reimagine conventional trading platforms as we know them. They can effectively crunch millions upon millions of data points in an instant, and make predictive models that are the most accurate according to the information,” he added.
Rahul Chari, CTO and cofounder, PhonePe shared how extensively PhonePe is using machine learning algorithms in a recent interview. The company uses a combination of data-driven nudges and user preferences. For example, for credit card bill repayment or electricity bill, a combination of ML and data science infers the bill and the due date at which the user’s reminder is set.
“Our machine learning model learns to personalise reminders on the basis of feed generated by the date of the bill and payment mode,” said Chari. Not only this, but PhonePe also used machine learning algorithms to drive increased conversion rates for users and partners, and curate personalised offers for users
Innoviti Payment Solutions runs a payments platform that has an ability to add intelligence to traditional payment channels, enhancing their value. Merchants, brands and financial service providers use these intelligent payment channels to reduce cost and drive sales of their products. In a media report recently, the company claimed that its machine learning (ML)-based Path Predictor technology has helped enable Google Pay across its stores. The combination of UPI through Google Pay and Path Predictor has led to volumes climbing to nearly 10% for several stores on the Innoviti network.
Disclaimer: This article aims to give a brief overview of machine learning applications in the finance sector and do not cover the technical aspects in detail. The information is sourced from primary and secondary sources.