Double Check Consulting has following Value Propositions to serve Clients honestly:-
- Insight Value: What others end up after analysis is insights but, we START with that insight. In other words, we start with penetrating hypotheses and right questions and prove or disprove them. So the money is saved and client’s knowledge about their business goes a few notches up. For example provider fraud hypotheses, “excess of zero paid dollars is indicative of fraud, waste and abuse” is very penetrating and saves millions of dollars in EACH of the specialties. This hypothesis can be used in retail, financial services and other industries. At present several service providers are promising to solve all the business problems of a company while they themselves are unable to find ONLY five hypotheses per week for their clients. How can a service provider solve all issues when they can’t find issues? Analytics, especially Big Data one is incremental value because it starts with what hypotheses/ use cases we know and such a promise results in companies not applying any good technology (https://insidebigdata.com/2017/08/28/dont-solve-wrong-big-data-problem/?cid=em-elq-20253&utm_source=elq&utm_medium=email&utm_campaign=20253&elq_cid=457076 ).
- Tool Value: To find hypotheses such as above-mentioned, we use our PRODUCT Multilevel Operational Learning Tool (MOL Tool); the mathematics for MOL Tool comes from calculus, algebra, trigonometry, Advance Statistics, Machine Learning Algorithms and Neural Network. Our MOL Tool is interactive, prompts to next steps in questions and hypotheses creation; talents don’t feel burdened by uncertainly of finding nothing. MOL Tool gives them plenty of opportunities to practice critical thinking, hence fear of failure is taken away. This helps us with result specific or MUST save or make money for clients. The greatest tool in the word such as R has no value to us because they do what we don’t want them to do. Double Check Consulting tool in E commerce fraud works in healthcare related use cases because Time Series Unsupervised method for outlier detection based on variables such as Charge dollar and paid dollar etc occurs in every industry. Similarly, our health care fraud tool works in other industry because it predicts character variables procedures, diagnosis, which we change to SKU_ID, store_ID etc. easily for retail and E- commerce use cases. In contrast to our technology, tools companies have doesn’t help with finding and/ or working with smart data (http://www.itproportal.com/features/can-retailers-and-manufacturers-turn-big-data-into-smart-data/?cid=em-elq-20253&utm_source=elq&utm_medium=email&utm_campaign=20253&elq_cid=457076 , https://spectrum.ieee.org/computing/it/the-future-of-big-data-distilling-less-knowledge-per-bit ).
- AI Value: Our AI strategy is to accommodate all variables for learning, be it 100 or 10K in a neural network. We learn about correlations in bulk and helps with finding hypotheses as mentioned above that works in multiple industries. We minimize model tuning to provide knowledge to clients on what the coding region in DNA or signals in each variable. For Healthcare fraud use cases we need to explain why certain claims are fraud or not. Combinations of such signals from correlations from a large number of variables simplify our analytics business process- we’re not perceived as black box methods. Our tools have depth! We’ve experienced by being hands on – that data-driven process breaks down during Fraud Predictive analytics process. That there needs to be a system to remove randomness in Business process. Several companies take one year to develop 500 variables for clients, 50 fraud conditions, and umpteen business rules. On Predictive analytics tools such as SAS Enterprise Miner or SPSS Predictive Modeler, you throw all that for feature selection and model tuning. If you remove one variable from the large pool of variables, entirely new set of variables are selected. One is forced to use 4-5 different predictive methods and build predictive models on variables that show up in most of the models. After doing the similar model tuning in each iteration while consuming considerable time, a maximum of 10 variables is selected in each of 4-6 iterations in one year. Knowledge concentrates on only those variables; resulting in large false positive and disconnect with ROI. While Predictive analytics rank Orders or technically correct, fraud is caught from low-value claims; other methods are more accurate and faster. When the model is presented to SIU they are surprised; many findings are contrary to their experiences. Hence the current Predictive models in 99% of companies are decision driven that comes short on Decision Making. They rather elucidate the limits of data. That’s because many service providers and their clients don’t have AI strategy (https://www.cio.com/article/3225445/artificial-intelligence/machine-learning-success-stories.html?utm_medium=nl&utm_source=internal&mrkid=11202185&mkt_tok=eyJpIjoiTURRd1pEWTBOVFZpTnpBMSIsInQiOiI2K1RIclNMaE9OK3BIa0Y0OVwvV3kzV1FLK2t6UEhkUnBUelVsYzdNS3VTNHlTUTdMOGN6Yk0rQTc2bkgrVkdUXC9mWXh1M3p3VFNHa1JwdXV0VGN1N0Z4WERWTTNuZXNXbUs5N2tvVFI5bnl1RnBlcXFTT3ZUU2w5cGcxRUdrektLIn0%3D ). Our AI strategy eliminates such bottlenecks; focus is on decision making not on spending large time experimenting with a large number of modeling methods. That’s why we know Broad data or data with more columns (variable) than rows is the smart data hiding within Big Data. We believe our methods removes model tuning randomness mentioned above by our AI strategy of accommodating all variables.
- Quality to Value: we perform “Hybrid Analytics”; outlier detection methodology and Scenario analysis to reduce false positives and false negatives for hypothesis such as, “as the zip code similarity increases, the propensity of billing certain procedures (CPT) increases from diseases external to the local environment”. For example, 9/11 created COPD center but you may find lots of claims in Tulsa, OK. Such an approach helps us with 30-60 days Fast Track POC with client’s data; so we define quality in terms of speed to delivery. While there is No tool on market for fraud use cases, our tool is not data hungry. Proven even in branded companies from United Health Group to Blue Shield of CA, It doesn’t depend upon the large volume of fresh data. We create variations in fraud data like one that happens in DNA Technology and find the moles before their scam squeezes the juice out of companies’ bottom line- Fast.
- Visualization Value: While BI data is governed, our brain is not. When companies slice and dice data, it starts producing visualizations to detriment – gaudy charts and graphs which generates lots of chatter but adds no value. Our AI produces Hypotheses in Bulk and when they’re used in visualizations, the business process is governed for ROI from end to end. There is no waste of time such as producing 100’s of charts from 200’s of water-cooler conversations.
- The effort to Value: Many hypotheses (insights) and questions never get acted upon even it has lots of business potential due of poor communications. Hence our efforts are in making sure what we find is acted upon. For example, neural nets and deep learning are “black box” solutions. With the help of our MOL tool, our talents make sure they are “light box” to business owners; they know as much as we do how we got to that result – getting them to act upon. The backbone here is data that has gone six sigma insights or visualization of hypotheses from data; another layer of AI on such a data helps with light box.
- Total Cost of Business Ownership: We keep the x-axis or Total costs lower while increasing the y-axis or revenue. We maximize the y-axis by finding nuggets of actionable information that client did not know before. For example, Predictive analytics tells us what will happen, not why it will happen. So when a decision maker looks at hypothesis presented above in visualization tool, s/he may find anyone whose claims are more than 10% zero paid dollars in a month, their claims have gone to clinical audit and proven fraud later 50% of the time. Now, this leads to a Predictive analysis use case” excess of zero paid dollar of 10% or more claims per month and clinical auditing on them 50% of the time is indicative of fraud, waste, and abuse.” What we’ve done is to know why Predictive analytics is going to find correlations. In other words, we have blended predictive analytics with prescriptive analytics or decision making. Prescriptive analytics will be performed by 10% of the companies by 2020- incredible differentiation. That means we’ve gone extra length for last miles of delivery for a successful project. We put ourselves on the steps of clients and are sensitive to failures that can harm our clients and hence Double Check Consulting (https://www.cio.com/article/3211485/project-management/why-it-projects-still-fail.html?idg_eid=2fdb8f1318827b225a160ef2a8c2b295&email_SHA1_lc=774b492232980b38ab5534483fab19ed4e0c0eb3&cid=ndr_nlt_Insider+Normal+Subscribers_2017-09-24&utm_source=Sailthru&utm_medium=email&utm_campaign=Best+of+Insider+2017-09-24&utm_term=Insider+Normal+Subscribers ).
- Talent Value: What’s a company without access to talents? We answered this question by mentoring 100+ talents since 2013 and when Double Check Consulting Facebook site started on 14th February 2014, 60 talents from India liked next day. They are there supporting us since then. On LinkedIn, Navin Sinha is connected to Top Talent houses from all over India and connected to 50 Data Scientists who are Engineers, Java, and Python programming experts and working on Deep Learning projects for FANG (Facebook, Apple, Netflix, Google) companies. They have expressed interest in working for Double Check consulting on more than one occasions. No one specializes in fraud Horizontal as Car insurance and Worker’s comp are clinically intensive; we’ve access to clinical coders, doctors, nurses. We’ve Billions of dollars worth clinical knowledge. Double Check Consulting is only BPO from India that is educated to American quality standard (MS in Tomato Genomics from University of Illinois and MBA in specialization in Decision Making from Utah State University), has its own fraud tools, proven work experience in fraud from multiple F500 companies, takes Fraud use case horizontally from healthcare, Car insurance, worker’s compensation, financial services, retail to complex multi-industry use case such as Anti-Money Laundry on the basis of access to diverse portfolio of talent. Insurance companies, for example, have lots of fraud issues but little domain expertise and access to talent that matters (https://www.itworldcanada.com/article/insurers-still-struggle-with-predictive-analytics-study/98415 ). Double check Consulting is certainly a help for them. That’s because while there are plenty of service providers to choose from, success comes from a business leader who is well grounded in technical concepts with robust hands-on experiences (https://thestack.com/big-data/2017/09/07/building-an-effective-big-data-strategy/?cid=em-elq-20253&utm_source=elq&utm_medium=email&utm_campaign=20253&elq_cid=457076 ).
- Curiosity Value: Our MOL Tool screens talent for curiosity, plus tool Educates talent into finding use cases from Disease Management, Healthcare Related Fraud, E commerce Fraud and Call Center Analytics – in a self learning and self creating environment.