Quriosity
THE
JANUARY 2013
FROM THE MENTOR’S DESK
2
FROM THE EDITOR’S DESK
3
MAIN STORYINSURANCE FRAUD PROTECTION MANOJ ARMANDLA QUANTS NEWS DIGESTABHISHEK KUMAR BOOK REVIEWCOMPETING ON ANALYTICS: VARUN S
4-7
8-9
10-11
QUANT GURU OF MONTH – D R KAPREKAR 12-13 BHAWNA JAIN
QUANTIZ OF THE MONTH: QUANTIZ TEAM
QUANTIZ FUN:
DEC 2012
14
VOLUME 3: ISSUE 9
From the Mentor’s Desk... Hi All, We are pleased to place the January issue of ‘Quriosity’ in your hands. Now the annual day is also approaching wherein we will be showcasing the Quants Talents through different activities. All are welcome to contribute with their ideas and make the annual day a grand success. Thanks for your interest in our activities. We are still expecting a lot more from the readers in the form of comments and suggestions. Thanks for your attention. Happy Reading! Regards Prof N.S.Nilakantan Mentor- QUANTINUUM
15
QURIOSITY
2
From The Editor’s Desk Editor’s note Greetings to all, It is the time when the Analytics is listed as the Top skill requirement for the Managers by the Leading Employment Magazines and Newspaper and here we are attempting to acknowledge you the importance of the same and give an insight that how it is going to shape the future. Keeping with this endeavor, we proudly present to you yet another special issue of “QURIOSITY – the ultimate quant magazine”. The cover story is our attempt to describe about the various analytics tools used to control the fraud in the institutes like Banks and Insurance along with the detail overview and description of the same. Quant digest furnishes interesting reads in the form of articles such as – “Mobile Application Launched for Retail Store Analytics”, “Taggstar unveils image analytics dashboard for publishers ”. The book review covers a book – “Competing On Analytics”, The New Science of Winning. Further to stimulate the gray matter in your brain; we bring you the wonders of our regular features- quant trivia, quantiz and quant fun.
Happy reading!! EDITOR
DEC 2012
QURIOSITY
3
INSURANCE FRAUD DETECTION If you’ve been used to thinking about Analytics in terms of sales or marketing, think again. Today, analytics can reinvent your enterprise technologies — social networking, big data, CRM — to crack down on financial offenders. Giving you more than an insight a day, to keep the fraud away. Digitization a new opportunity for fraud detection? Digitization marked by a growing number of mobile devices and social media is changing the business landscape for all sectors — including insurance. The opportunities offered by this landscape for insurers are vast. Social networks and communities help insurers connect with their customers better, which in turn aids branding, customer acquisition, and retention. Insurance firms also receive a plethora of inputs from digital information in the form of feedback, which also can be used to come up with customized products and competitive pricing. In addition to these opportunities, insurance companies are harnessing digitization — using data analytics for fraud detection. Handling fraud manually has always been costly for insurance companies, even if one or two low incidences of high-value fraud went undetected. In addition to this, the big data trend, (the growth in unstructured data) always leaves lot of room for a fraud going undetected if data is not analyzed thoroughly. Traditionally, insurance companies use statistical models to identify fraudulent claims.
These models have their own disadvantages. First, they use sampling methods to analyze Fraud DEC 2012
QURIOSITY
4
Equity Research: Equity research is analysis of equity security of companies or industries. Financial services KPO’s offer two kind of roles in Equity re-search namely sell-side & buy-side. In sell-side, the analyst works with stock analyst from a brokerage firm. In buy side the analyst typically works with portfolio
data, which leads to one or more frauds going undetected. There is a penalty for not analyzing all the data. Second, this method relies on the previously existing fraud cases, so every time a new fraud occurs, insurance companies have to bear the consequences of the first time. Finally, the traditional method works in silos and is not quite capable of handling insurance the evergrowing sources of information from different channels and different functions in an integrated way. Analytics addresses these challenges and plays a very crucial role in fraud detection for insurance companies. Some of the key benefits of using analytics in fraud detection are discussed below. Identification of Low Incidence events: Using sampling techniques comes with its own set of accepted errors. By using analytics, insurance companies can build systems that run through all critical data. This in turn helps detect low-incidence (0.001%) events. techniques such as predictive modeling can be used to thoroughly analyze instances of fraud, filter obvious cases, and refer low-incidence fraud cases for further analysis. Enterprise Wide Solution: Analytics help in building a truly global perspective of the anti-fraud efforts throughout the enterprise. Such a perspective often leads to effective fraud detection by linking associated information within the organization. Fraud can occur at a number of source points: claims or surrender, premium, application, employee-related or third-party fraud. At the same time, insurance channel diversification is adding to the fragmentation of traceable data. Insurance-related activities can be done via mobile devices apart from the traditional online and face-to-face insurance. This can be viewed as an addition to information silos in the insurance industry. Given greater channel diversification and the increase in areas where fraud can occur, it is important for insurers to have accessible enterprise-level information about their business and customers.. Data Integration: Analytics plays an important role in integrating data. Effective fraud detection capabilities can be built by combining data from various sources. Analytics also help in integrating internal data with third-party data that may have predictive value, such as public records. Data sources with derogatory attributes are all public records that can be integrated into a model. Examples include bankruptcies, liens, judgements, criminal records, foreclosures, or even address change velocity to indicate transient behavior. Other types of third-party data can be beneficial in enhancing efficiencies such as review of appraisal information to determine if damages match description or loss or injuries being claimed. One of the most under-utilized data sources is medical bill review data. This data, if used in a model properly, is a gold mine for companies investigating medical fraud. Uncovering anomalies, in billing and adding these to the other scoring engines or social network analysis will decrease the amount of time an investigator or analyst spends trying to pull all of the pieces together to identify fraudulent activity.
DEC 2012
QURIOSITY
5
A sample innovative fraud detection method: Social Network Analysis (SNA) Let’s take an example to explain the use of social network analysis (SNA). In a car accident, all people in the vehicle have exchanged addresses and phone numbers and provided them to the insurer. However, the address given by one of the accident victims may have many claims or the driven vehicle may have been involved in other claims. Having the ability to cull this information saves time and gives the insurer an insight into the parameters involved in the fraud case. SNA allows the company to proactively look through large amounts of data to show relationships via links and nodes. The SNA tool combines a hybrid approach of analytical methods. The hybrid approach includes organizational business rules, statistical methods, pattern analysis, and network linkage analysis to really uncover the large amounts of data to show relationships via links. When one looks for fraud in a link analysis, one looks for clusters and how those clusters link to other clusters. Public records such as judgments, foreclosures, criminal records, address change frequency, and bankruptcies are all data sources that can be integrated into a model. Using the hybrid approach, the insurer can rate these claims. If the rating is high, it indicates that the claim is fraudulent. This may be because of a known bad address or suspicious provider or vehicle in many accidents with multiple carriers. SNA follows this path: Insurance fraud detection using social network analysis is as follows:
Before implementing SNA, insurers should consider: 1. How fast data arrives DEC 2012
QURIOSITY
6
2. How clean the data is when it arrives 3. How deep the analysis must go to get the results 4. What type of user interface components need to be included in the SNA dashboard Real time Application: GE Consumer & Industrial Home Services Division Scenario In GE Consumer & Industrial Home Services Division, claims typically came from technicians who repair consumer products that are under warranty. One of the biggest problems with their old process was that they could not identify patterns. With the amount of data available to them, no one could see unusual behavior emerging. Sometime back, GE got the perfect scenario to test an SNA solution from SAS, a developer of business analytics software. The company was tipped off to some service providers committing fraud. This situation made for an ideal pilot scenario. SAS was given the responsibility of analyzing the available data and identifying patterns in the data to find out who was committing the fraud. Functioning of the fraud detection system Typically, there are some metrics and indicators on every claim that assist in identifying suspicious or fraudulent claims. GE’s claims data is fed into the fraud detection software. There are 26 claim-level analyses, which are automatically calculated for each claim. There are some indicators like flags that are calculated based on various metrics and sent for auditing when they indicate that multiple elements in the claim fall out of the normal curve. Once these claims are flagged, the auditors at GE investigate these suspicious claims. Outcome The GE Consumer & Industrial Home Services Division estimated that it saved about $5.1 million in the first year of using SAS, to detect suspect claims.
Manoj Aramandla PGDM-FS (2012-14)
QUANTS NEWS DIGEST “Mobile Application Launched to enable Retail Store Analytics” The patent-pending technology employed in the app aims to enable retail store owners and managers to start collecting data and receiving Google Analytics-type analytics directly on their own mobile devices, without the need to use any additional in-store hardware. It will present visitor analytics such as number of visitors, dwell time and visiting loyalty as well as window conversion rate in real-time. RapidBlue COO Sampo Parkkinen (pictured) said: ”Until today, offline retail store analytics has been produced by using data collected by in-store hardware, either proprietary or for instance existing WiFi-hardware. Using hardware to produce analytics aways encompasses a cost of provision. Due to this cost, certain limitations as to who can actually use offline retail store analytics have been present in the market. Our Mobile Retail Store Analytics changes that. By enabling retail store owners to simply download an application to receive store visitor analytics, we open up the entire tail-end retail market, for whom any type of hardware solution is not an alternative.” The application is currently available for the iPhone, iPad, Android and Meego to interested beta-testers.
“Taggstar unveils image analytics dashboard for publishers” The new analytics tool lets publishers see which images are driving the highest levels of traffic and engagement. Data points within the dashboard are available in real-time. Metrics featured on the dashboard include unique views, dwell time, the images that have been shared the most via social platforms, trackback image data via social platforms such as Facebook and image view data over a period of time. The aim of the service is to help publishers think more strategically about the images that they are using on their site, and begin tracking how well their images are performing. The dashboard is free to use.
OE D CC T 22 00 11 22
QURIOSITY
8
Fraser Robinson, founder and CEO of Taggstar, said: “By monitoring the performance of images via our dashboard, users can make better, more informed decisions about the sort of images they are using, and where they are using them on their site. Image analytics and data are a key part of Taggstar’s core proposition. Images are typically the most powerful and engaging piece of content on a page, and our goal is to help publishers bring the value of their image assets into focus.”. http://www.research-live.com/
Abhishek Kumar PGDM (2012-14)
DEC 2012
QURIOSITY
9
BOOK REVIEW: COMPETING ON ANALYTICS Competing on Analytics, is authored by Tom Davenport and Jeanne Harris and discusses a number of (what they call) "analytic competitors," that is to say companies that use their analytic prowess not just to enhance their operations but as their lead competitive differentiator. The authors' formal definition of an analytic competitor is, "The extensive use of data, statistical and quantitative analysis, explanatory and predictive models, and fact-based management to drive decisions and actions." The book has two parts - one on the nature of analytical competition and another on building an analytic competency. The first part describes how analytics are used in both internal and external processes. The second part lays out a roadmap that describes how an organization can become an analytical competitor and how to manage analytical people. This section also provides a quick overview of a business intelligence architecture and offers some predictions for what the future of analytics holds. The authors argue that organizations that use analytics extensively and systematically are able to outthink and out execute their competition. They also argue that organizations must have a strategic distinctive competency. Without one, they cannot be an analytic competitor. Their experience suggests, however, that analytical competitors may start with a primary focus, but the culture of testanalyze-learn spreads quickly and widely. To be successful, analysis has to be an overarching company skill and not just the province of a few rocket scientists. In addition, the book outlines what they call four pillars of analytical competition - a distinctive capability, enterprise-wide analytics, senior management commitment and large-scale ambition - and also lays out five stages of analytic competition from "analytically impaired" to" analytic competitor". The importance of experimentation is made clear, and the book repeatedly emphasizes the need for companies and executives to be willing to run their business "by the numbers." The book is full of stories about how companies compete analytically, for example: 


Capital One's focus on identifying and serving new market segments before its peers can. They have a lovely concept of "deaveraging" - breaking a segment into small segments for better targeting. Marriott's total hotel optimization using a new measure called "revenue opportunity" - what percentage of the theoretical maximum revenue they actually made. They got this to rise an amazing 8 percent. Progressive Insurance is so certain that if another company offers you a better rate then you would not be profitable and so are willing to disclose what their competitors' rates are.
DEC 2012
QURIOSITY
10
The Veteran Administration's use of evidence-based medicine and predictive analytics, along with automated decisions for treatment protocols, is noted, as is the fact that perhaps only 25 to 30 percent of medical decisions are scientifically based. Honda makes good use of text analytics to flag early problems in cars by analyzing warranty claims calls by customers or dealers to headquarters. Vertex, a pharmaceutical company, starts by identifying the right metric to measure success and then drives into the data needed to measure that. Harrah's focuses on real-time analytics at the point of sale so that action can be taken as it is being collected. DnB NOR bank uses event triggers to prompt customer relationship offers, using analytics to trigger the right events. O2, a mobile phone company, uses personalized menus to maximize value of limited phone real estate and uses predictive analytics to add personalization. CEMEX used analytics to move focus from the sale of a commodity (cement) to the delivery window using analytics and GPS. They went from three hours for a change to 20 minutes. Netflix focuses on giving each customer a personalized website experience based on recommendations, ratings and segmentation.
The stories illustrate many factors, from creating new measures to tracking the right measures and from the need to change your perspective to the power of executive sponsorship. The book also has a great list of questions regarding new initiatives - how will it make you more competitive, what data do you need, does the technology work and what complementary changes need to be made in order to take full advantage of these new capabilities. They outline a number of ways to get a competitive advantage from data - by collecting unique data, manipulating data better, using a unique algorithm or embedding it in unique process. Regardless of the competitive approach, the need for analytical executives' willingness to act on the results of analysis was clear: segmentation of customers is not enough...you must differentiate their treatment to make a difference. Concluding, they outline three options - give them powerful analytic/data mining tools, have the system spit out the "right" answer or give them a spreadsheet. They think there is a fourth option automate the decision, but have it return many options that provide reasoning and visualization of the analytics. This does not require the automation to be "perfect," but it does help the individual make better decisions. The individual, in this scenario, is not wading through the whole report, but instead is looking at three to four snippets each focused around a specific treatment option. Decision automation need not be all or nothing. VARUN S PGDM-FS 2012-14 DEC 2012
QURIOSITY
11
QUANT GURU of the MONTH D R Kaprekar was born in Dahanu, a town on the west coast of India about 100 km north of Mumbai. He was brought up by his father after his mother died when he was eight years old. His father was a clerk who was fascinated by astrology. Although astrology requires no deep mathematics, it does require a considerable ability to calculate with numbers, and Kaprekar's father certainly gave his son a love of calculating. Kaprekar attended secondary school in Thane, which is northeast of Mumbai. There he spent many happy hours solving mathematical puzzles. He began his tertiary studies at Fergusson College in Pune in 1923. There he excelled, winning the Wrangler R P Paranjpe Mathematical Prize in 1927. This prize was awarded for the best original mathematics produced by a student and it is certainly fitting that Kaprekar won this prize as he always showed great originality in the number theoretic questions he thought up. He graduated with a B.Sc. from the College in 1929 and in the same year he was appointed as a school teacher of mathematics in Devlali, a town very close to Nashik which is about 100 km due east of Dahanu, the town of his birth. He spent his whole career teaching in Devlali until he retired at the age of 58 in 1962. Kaprekar did manage to publish some of his ideas in low level mathematics journals, but other papers were privately published as pamphlets with inscriptions such as Privately printed, Devlali or Published by the author, Khareswada, Devlali, India. Kaprekar's name today is well-known and many mathematicians have found themselves intrigued by the ideas about numbers which Kaprekar found so addictive. Following are some of the ideas which he introduced. Kaprekar number A Kaprekar number is a positive integer with the property that if it is squared, then its representation can be partitioned into two positive integer parts whose sum is equal to the original number (e.g. 45, since 452=2025, and 20+25=45, also 9, 55, 99 etc.) However, note the restriction that the two numbers are positive; for example, 100 is not a Kaprekar number even though 1002=10000, and 100+00 = 100. This operation, of taking the rightmost digits of a square, and adding it to the integer formed by the leftmost digits, is known as the Kaprekar operation.
DEC 2012
QURIOSITY
12
Kaprekar’s Constant One starts with any four-digit number, not all the digits being equal. Suppose we choose 4637. Rearrange the digits to form the largest and smallest numbers with these digits, namely 7643 and 3467, and subtract the smaller from the larger to obtain 4167. Continue the process with this number subtract 1467 from 7641 and we obtain 6174, Kaprekar's constant. Let’s try again. Choose 3743 7433 8640 8721 7443 9963 6642 7641
-
3347 0468 1278 3447 3699 2466 1467
= = = = = = =
4086 8172 7443 3996 6264 4176 6174
Again we have obtained Kaprekar's constant. In fact applying Kaprekar's process to almost any fourdigit number will result in 6174 after at most 7 steps. This was first discovered by Kaprekar in 1946 and he announced it at the Madras Mathematical Conference in 1949. He published the result in the paper Problems involving reversal of digits in Scripta Mathematica in 1953. Clearly starting with 1111 will yield 0 from Kaprekar's process. In fact the Kaprekar process will yield either 0 or 6174. Exactly 77 four digit numbers stabilize to 0 under the Kaprekar process, the remainder will stabilize to 6174. Anyone interested could experiment with numbers with more than 4 digits and see if they stabilise to a single number (other than 0). Devlali or Self number In 1963, Kaprekar defined the property which has come to be known as self numbers, which are integers that cannot be generated by taking some other number and adding its own digits to it. For example, 21 is not a self number, since it can be generated from 15: 15 + 1 + 5 = 21. But 20 is a self number, since it cannot be generated from any other integer. He also gave a test for verifying this property in any number. These are sometimes referred to as Devlali numbers (after the town where he lived). Demlo number Kaprekar also studied the Demlo numbers, named after a train station where he had the idea of studying them. These are the numbers 1, 121, 12321, …. which are the squares of the repunits 1, 11, 111, …. BHAWNA JAIN PGDM- FS 2012-14 DEC 2012
QURIOSITY
13
QUANTIZ of the MONTH Q1) Gold is 19 times heavy as water and copper 9 times as heavy as water. What is the ratio in which these two metals be mixed so that the mixture is 15 times as heavy as water? Q2) If the compound interest on a certain sum for 2 years at 3% is Rs. 101.50, what will be the corresponding simple interest? Q3) A train 300 meters long is running at a speed at 250meters per second. In how much time it will cross a bridge of 200 meters long? Q4) In 5 compulsory subjects a candidate secures an average of 60%. In 2 optional subjects he secures equal marks and when these are added to his total, his average drops by 4%. How much did he secure in each optional subjects? Q5) A candidate scores an aggregate of 60% marks, scoring an average of 56% in 4 of the papers and 68% in the others. How many papers were there totally?
Please send us the answers at simsr.quantinuum@gmail.com. Answers and Name of the winner (first all correct /most correct entry) will be published in the next issue. Solutions to last issue’s Quiz of the month 1. He was born on December 31st and spoke about it on January 1st 2. 100 = 177-77 = (7+7)x(7+(1:7)) 3. Sum of all numerals must be ten because each numeral stands for the count of other numerals and because this number shall have ten numerals. Beginning to choose reasonable numerals for the first figure you can come across the correct number: 6210001000. 4. 102564 5. There were initially 190 balls in the urn; 10 each of 19 different colors.
Keeping the Grey matter Alive! QUANTIZ TEAM DEC 2012
QURIOSITY
14
QUANT FUN Sudoku of the Month.
QUANT TRIVIA
Please send us the answers at simsr.quantinuum@gmail.com. Answer and name of the winner will be published in the next issue.
2
3
1
5
8
6
4
7
9
5 4 9 3 1 7 8 6
8 6 4 2 7 1 5 9
9 7 5 6 8 2 3 4
1 2 3 8 6 4 9 7
7 3 1 4 9 6 2 5
4 9 7 5 2 8 1 3
2 1 6 9 3 5 7 8
3 5 8 1 4 9 6 2
6 8 2 7 5 3 4 1
DEC 2012
QURIOSITY
“A Mathematician is a blind man in dark room looking for black cat which is not there. – Charles Darwin.”
15
QuantConnect Quantinuum, the Quant's forum of KJ Somaiya Institute of Management Studies and Research is formed with two objectives. Firstly to remove the common myth from the students mind that mathematics is difficult. Secondly to give students an exposure on how to make decisions in real life business problems using quantitative techniques. This helps to bridge the gap between theory and the practical application. For any further queries and feedback, please contact the following address KJ Somaiya Institute of Management Studies and Research Vidya Nagar, VidyaVihar, Ghatkopar East Mumbai -400077 Mentor Prof. N.S.Nilakantan (9820680741) – email – nilakantan@simsr.somaiya.edu Team Leaders Satyadev Kalra (8291687568) Abin Abraham (9594374903) Somjeet Dutta (9769513003) Vaibhav Goel (9769456493) Editorial Team Aditi Paliwal (9819116068) Abhishek Kumar (9819099671) Manish Murthy (9167679676) For any queries, drop us a mail at simsr.quantinuum@gmail.com For more details: http://quantinuum.weebly.com/
Like our newsletter? Want to contribute and see your name in print? Rush your articles, concepts, trivia, facts and news about the Wonderful World of Numbers to us by email to simsr.quantinuum@gmail.com. simsr.quantinuum@gmail.com