Find It. Define It. Fix It.
Thanks for reading our first article in the RIViR Reads series. In RIViR Reads, we’ll be exploring the exciting and constantly evolving world of data and information science. We’ll examine how companies are developing innovative and creative technologies in manipulating data, and learn about clever techniques to solve the world’s problems.
Scenario: A physician serves patients for a large, regional health system. This provider sees patients at an office in the Capital City Monday through Wednesday, and travels an hour and a half to see patients visiting the regional health provider’s office in a small town. The patients’ treatments are virtually the same in each location, except for the few extra padded billing codes in the rural area since many of these small-town patients are on Medicaid.
Scenario: A stock trader works at a medium-sized brokerage in Chicago with a knack for knowing precisely when to short the right stock. The trader doesn’t deliver stock advice to the brokerage’s customers or in any data feeds. Instead, trades are brokered using a shell account held by the trader’s college roommate’s niece; ten years old.
Scenario: A once promising freshman, now sophomore is up late. The student is burning the midnight oil at a large, mid-Atlantic state university struggling to comprehend the Greek symbols in their Calculus II textbook. Having already withdrawn from two other classes, the student is afraid of flunking out by failing one more Calc II exam.
Each one of these three scenarios are different. Each has a different set of players, using different methods, and have different outcomes. Yet, each scenario shares a common characteristic. They are all anomalies as defined in the systems their players operate in.
Anomalous behaviors and bad actors have a real cost. Fraudulent activity and theft cost US taxpayers billions of dollars every year in real economic value, as well as realized human costs when resources could be used more effectively serving patients in communities and populations who need it most. What often goes unreported, but is just as impactful, is the real social cost as anomalous behaviors and reversible actions go unchecked and unrealized potential never reaches society.
“We accumulated so much data that we didn’t know how to store it, let alone how to actually use it.”
As systems have become more complex and harder to understand, organizations threw increasing amounts of data at the problem – the idea being that if we had more information, we could take action and solve our problems. We accumulated so much data that we didn’t know how to store it, let alone how to actually use it. New data warehousing procedures and concepts revolving around new schemas and data strategies led to the culmination of the first visualization products, advertising: “if you could see what the data looked like you could make the necessary decisions and take action on your business problems.” We were able to get data out of spreadsheets and pivot tables into snazzy applications. We were able to see trends in charts, scatter plots showing multiple data points at once, and graphs to connect the dots.
Gigabytes of data, pretty pictures, and all of the linear regressions in the world still made it hard to find anomalies. Cheap GPUs and even cheaper memory made machine learning (ML) a practical solution for running statistical and predictive models used for anomaly detection. The glut of accumulated data made decision trees and nearest neighbor techniques painfully slow for modern uses. Furthermore, the bad guys would do just enough to stay under the radar, becoming more sophisticated and adapting to policy shifts in our systems. Artificial Intelligence, provided by state-of-the-art convolutional neural networks made it possible to train algorithms against millions of data points while significantly increasing anomaly detection performance. AI algorithms, written in the new Python programming language provided a much-needed key for simplifying the code needed to load data in memory and write model code to identify anomalies quickly.
Visualization products, a new language, and innovative neural networks jumpstarted the business intelligence software industry. BI tools like Tableau and PowerBI, fed with data generated by AI/ML algorithms made it easier to find things. BI brought us much needed insight, but it brought us no closer to taking action on our business problems.
The stage is set, and now enters RIViR.
“Give me a list of things, and help me work down that list.”
We can collect all of the data in the world; and look at it all day, but until we make decisions and take action on that data, we will not improve.
Where BI tools stop at the edge of identifying and visualizing data, RIViR takes BI one step further by enabling organizations to take definable, traceable actions on their challenges and measure their effectiveness in one tool.
Qlarant’s Risk Identification, Visualization, and Resolution solution platform is built on open standards, powered by the latest AI/ML kits, and empowers business to build applications that help find their most challenging problems, see impacts on their organization, and resolve issues in a comprehensive platform.
RIViR delivers value by enabling organizations to define the application they need to solve their business problems by treating algorithms, visualizations, and business processes as authored content. Organizations know their business better than anyone, and RIViR provides the tools and platforms to build several solutions.
RIViR solutions are built using a powerful, new type of algorithm created by Qlarant – Nerdy Algorithms. These are specialized algorithms that know how to visualize anomalous data, identify anomalies, and know how to fix issues caused by them. When Nerdy Algorithms are fed data, their underlying models deliver the most important anomalies to the user and show the impacts those anomalies are making through a series of charts and graphs displaying resultant data. Then, if the user wants to take corrective action on any particular outlier, the Nerdy Algorithm recommends the most appropriate course of action the user should take.
The Risk Identification (RI) module provides users with tools they can use for anomaly detection, searching, and reviewing data. The RI module makes referenced and observable data available at a glance for the user, and gives users the option to deep dive into specific supplemental information for detected anomalies.
The Risk Visualization (Vi) module gives advanced visualization capabilities. RIViR provides the ability to deliver custom dashboards to different user types, and makes it easy to retrieve companion visualizations for any anomaly.
The Risk Resolution (R) module is special to RIViR. Other BI tools may detect anomalies, present visuals, or possibly do both in a clumsy way. The R module gives users the power to take corrective actions by following a directed, business tailored series of steps that can be performed by the user. Additionally, RIViR gives organizations the tools to schedule, assign, and monitor productivity while users are performing their duties using Courses of Action.
This is an exciting time to be in the anomaly detection business. We are collecting more and more information every day, and our globally-connected world adds increasing complexity. We’ve yet to crack the code on moving away from information hoarding to realizing all of the untapped potential hidden away in our data. Solutions made by Qlarant and others are taking new and exciting paths in the utility of data.
RIViR can find the problem, define it, and help you fix it.