LATEST UPDATES
Card-image-cap

Technology | Tech Start Ups

AI for Credit Scoring – An Overview of Startups and Innovation

Dec 13, 2018   •   by   •   Source: Proshare   •   eye-icon 8710 views

Thursday, December 13, 2018      09.25AM / By Niccolo Mejia of Emero

 

Severalcompanies offer AI-based credit scoring applications to banksand enterprise creditors looking to better understand the risk associated withtheir potential borrowers. Traditional methods of credit scoring take intoconsideration the credit histories of potential borrows, but this might notallow certain people access to credit despite the fact that they could paytheir loans back when their payments are due.

 

AIcould allow banks and creditors to score potential borrowers on theircreditworthiness using alternative data, specifically that from social mediaposts and Internet activity: what sites someone visits and what they purchasefrom eCommerce stores. Online behavior can indicate whether a person is likelyto pay back their loans, and AI could allow banks and creditors to factor thisinto their assessments of their potential borrowers.

 

What Business Leadersin Banking Should Know

Artificialintelligence solutions for credit scoring more often than not are predictiveanalytics solutions. This makes sense given that credit scores are, in effect,scores predicting the likelihood that a customer will pay back their loans.

 

Generallyspeaking, these vendors offer software that allow banks to minethe web for information on potential customers. The software might factor in acustomer’s social media posts or the sites on which they’re an active user intoits assessment of the customer’s creditworthiness. The machine learning modelwould need to have been trained on labeled datasets indicating which kinds ofsocial media posts or websites are indicative of a responsible customer andwhich are indicative of a risky customer. In order to come up with a score, thesoftware runs all of the information it collects on the customer through itsalgorithm and calculates if the bank would be taking a small or large risk ifit underwrotethem.

 

This, insome cases, could provide people access to credit who wouldn’t have access toit through traditional means. Credit scores require a credit history, but manypeople without credit histories would be able to pay back their loans givenaccess to credit. Banks often don’t feel comfortable lending to them, however,because they don’t have credit histories.

 

Again,however, some AI credit scoring software may work differently, and eachsoftware likely weighs certain online behaviors differently. For example, somecompanies tout their software’s naturallanguage processing abilities. Although all of the companies in thisreport offer software that seem to use some amount of natural languageprocessing to mine the social media of its potential borrowers’, some companiesare more upfront about it than others.  At the very least, it seems thesecompanies consider the capability worthy of the spotlight in its marketingmaterials.

 

Thecompanies that are selling AI credit scoring software or using it internallyall seem to have the requisite talent for building and managing machinelearning software, at least to some degree. This density of talent isrelatively rare, especially in industries like marketing.There are also numerous vendors selling purported AI software for variousapplications in finance,but these companies are often lacking in the kind of AI talent we look for whenvettinga company on their claims of leveraging AI.

 

Of thecompanies in this report, ZestFinance and Kreditech seem to have the highestdensity of data science talent. These data scientists holds Master’s and PhDsin computer science and various statistical fields and one in artificialintelligence itself. Also, Kreitech employed a ChiefData Officer with a PhD in AI from 2014 to 2018. As a result, webelieve these companies have a very high likelihood of leveraging genuinemachine learning software. This bodes well for them. Kreditech uses theirsoftware internally, but ZestFinance offers their software to enterprise banks.

 

LendoEFLemploys a CTOwith a PhD in Neuroscience from 2002, and the company also has several datascientists on its team. Two of these data scientists hold advanced degrees (aMaster’s and a PhD), but neither in computer science. Despite, we thinkLendoEFL has a relatively high chance of actually offering a machine learningsolution. Their company is not as robust as ZestFinance and Kreditech when itcomes to AI talent, but they’re likely not lyingabout doing AI either.

 

SAS isof course the most established enterprise amongst the companies in this report,and although they certainly have the money to employ data science talent, wecaution readers about automatically trusting large companies when they claim tooffer AI. Many companies will hire data scientists just to say they areleveraging AI. They then fail to provide their data scientists with anymeaningful work because they lack the ability to speak to them in terms theycan understand. In other words, their subject-matterexperts and data scientists don’t and can’t communicate well, andthis is imperative for building machine learning software. The techgiants like Google, Facebook, and Amazon see success in their AIendeavors in large part because even those employees that aren’t data orcomputer scientists at those companies know how to speak the language in a waythat allows them to inform the AI products the data scientists develop.

 

 

Proshare Nigeria Pvt. Ltd.

LenddoEFL offers asoftware called LenddoScore. We previously discussed Lenddo in our report on NaturalLanguage Processing in Finance. The company claims its softwarecan help banks determine the creditworthiness of potential borrowers usingpredictive analytics and natural language processing. Lenddo advertisesthat their software can be used to reach prospects that are unable to get creditbecause their software factors in data from online activity as opposed tocredit history.

 

LenddoEFLclaims users can install the company’s application onto their smartphones. Thesoftware uses natural language processing to analyze users’ social media postsand what they type into their browser for indicators of responsibility orrisk-taking. Then, this information informs the predictive analytics algorithmthat creates a credit score out of it. Banks and credit unions can then use theusers’ LenddoScores to better understand the risk they pose of not paying backtheir loans.

 

We caninfer the machine learning model behind the software was trained on thousandsof customer data points including social media posts and internet browsingbehavior. This data would have been labeled as indicative ofresponsibility or risk from the perspective of banks and creditors. The datawould then be run through the software’s machine learning algorithm. This wouldhave trained the algorithm to discern which data points correlate tocreditworthiness. The software would then be able to predict whichprospects are most likely to pay back their loans.

 

Belowis a short 2-minute video demonstratinghow LenddoEFL’s software works:

 

LenddoEFLdoes not make available any case studies reporting success with their software.Also, we were unable to find any mention of enterprise-level companies onLenddoEFL’s website nor in any of their press releases, but they have raised$14 Million and are backed by Golden Gate Ventures.

 

NaveenAgnihotri is CTO at LenddoEFL. He holds a PhD in Neuroscience fromColumbia University. Previously, Agnihotri served as CTO at Milabra.

 

Proshare Nigeria Pvt. Ltd.

ZestFinance offers a softwarecalled ZAML, which it claims can help financial agencies determine thecreditworthiness of potential borrowers and reduce loan defaults usingpredictive analytics and natural language processing.

 

We caninfer the machine learning model behind the software was trained on thousands ofsocial media posts, geolocations, browsing activities, and other data points.This data would have been labeled as indicative of risk or not from a bank orcreditor’s perspective. The data would then be run through the software’smachine learning algorithm. This would have trained the algorithm to discernwhich data points correlate to the types of borrowers that are most and leastlikely to pay back their loans.

 

Intheory, the software would then be able to predict if loan applicants arelikely to pay back their loans or not. That said, we could not find avideo demonstrating how the software works explicitly.

 

ZestFinanceclaimsto have helped Prestige Financial Services reduce lending losses anddefaults without sacrificing credit approval ratings. Prestige usedZestFinance’s software to complement their traditional underwriting methods.According to the case study, Prestige saw a 33% decline in credit losses, alongwith a 14% increase in borrower approval ratings. ZestFinance also listsFord Credit as one of their past clients.

 

Jay Budzik is CTO atZestFinance. He holds a PhD in Computer Science from Northwestern University.Previously, Budzik served as Chief Product Officer at Kinetic Social.

 

 

Proshare Nigeria Pvt. Ltd.

Kreditechuses a namesake software internally, and it helps the company determinethe creditworthiness of potential borrowers who do not have an extensivebanking history using predictive analytics and likely natural languageprocessing.

 

We caninfer the machine learning model behind the software was trained on hundreds ofthousands of customer data points from social media and internet browsinghistories. These would involve conversations about monetary exchanges, as wellas data points from eCommerce sites and payment processing sites, such asAmazon and Paypal respectively. This data would have been labeled as positiveor negative indicators of responsibility and creditworthiness. The labeled datawould then be run through the software’s machine learning algorithm. This wouldhave trained the algorithm to discern the data that correlates to andcreditworthiness.

A usercould then feed the software with a potential borrower’s social media posts,for example, and the algorithm would search it for indicators of responsibilityor creditworthiness. For instance, the software could comb through aconversation on the potnetial borrower’s social media and find that theypromised a coworker that they would be paid back by a certain date. Thealgorithm might then label this activity as a positive indicator ofresponsibility.

 

Kreditechdoes not list any major companies as clients, but they have raised $497.3Million and are backed by Rakuten and J.C. Flowers and Co.

 

 

Proshare Nigeria Pvt. Ltd.

SASoffers a software called CreditScoring for SAS Enterprise Miner, which it claims can help banks andfinancial agencies predict credit risk using predictive analytics.

 

We caninfer the machine learning model behind the software was trained on thousandsof borrower profiles and credit histories. The data would then be run throughthe software’s machine learning algorithm. This would have trained thealgorithm to discern which data points correlate to a borrower who poses ahigher risk to the company than others. The software would be able topredict the amount of risk associated with a potential borrower.

 

Wecould not find a demonstration video available for the software.

 

SASInstitute claimsto have helped Piraeus Bank Group speed up data analysis and reportgeneration. Piraeus Bank Group integrated SAS Institute’s software into itscore banking system so that it could access their data. According to the casestudy, Piraeus Bank Group was able to improve data analysis speed by 30%. SASInstitute also lists Bank of America and Honda as some of its past clients.

 

Jim Goodnight is CEO atSAS Institute. He holds a PhD in Statistics from North Carolina StateUniversity. Goodnight has spent the last 42 years of his career at SASInstitute.

 

 

For Further Reading

AIin Banking – An Analysis of America’s 7 Top Banks

 

Proshare Nigeria Pvt. Ltd.


Related News

1.        How To Build BigThings With A Small Team

2.       NaspersAnnounces $300M Initiative To Support Startups And Tech In South Africa

3.       After AnnouncingA $1.5M Seed Round This Year, Uganda-Based Asaak Is Raising Again

4.       Here Is How AMobile Punch Dashboard Can Help You Track Your Employees’ Overtime

5.       LiquidationPreference: Your Start Up Equity Could Be Worth Millions—Or Nothing

6.       World's FirstShared Trading Platform, Exchain, Launches Public Beta Test

7.       10 TechnologiesThat Will Disrupt Banking in Next 5 Years

8.       The Biggest SoftwareAcquisition In History

9.       The 7 MostAffordable US-Based Web Hosts

10.    Natural LanguageProcessing in Banking – Current Applications

11.     TheTransformation Of The Traditional Finance Function

12.    FacebookLaunches Video-call Device ‘Portal’ Amid Data Privacy Concerns

13.    High Speed 5GTechnology Revolution In Africa – Vodacom Picks Lesotho As Testbed

14.    Nigerian OnlineTrading Portals Ranking Report:5 Takeaways

15.     The SingleBiggest Breakthrough In Oil Tech This Year

16.    The NewlyRedesigned Proshare Market: 8 Things to Note

17.     Machine Learningin Finance – Present and Future Applicati...

18.    IdentityManagement, Payment Service Bank Key to Scaling Financial Inclusion in Nigeria

19.    The DifferenceBetween Artificial Intelligence and Machine Learning

20.   AI in ForeignExchange Trading (Forex) – Current State of the Sector

21.    #NES24 -Breaking the Financial Inclusion Barrier

22.   Digitization IsGaining Momentum In the Capital Market Ecosystem-Oscar Onyema, CEO NSE

 

Proshare Nigeria Pvt. Ltd.

Get the App

apple-store  play-store

Connect with us


Proshare is a professional practice focused on delivering research and information services to bridge the gap between investors and markets; by delivery on credible, reliable, and timely engagements through the following areas — Impact Research, Market Intelligence, Strategic Advisory, Stakeholder Relations & Digital Media.