Analytics Services

Analytics Services

Challenging problems. Flexible analytics. Superior results.

Survey Analysis Solutions

Understanding the voice of the customer

Surveys help an organization learn key issues and improvements suggested by customers and employees. To better understand the results of a survey, an organization needs to go beyond structured questions and analyze text responses to open ended questions. Our Group provides a state-of-the-art survey analysis tool for thorough analysis and automated coding of text responses, combined with the powerful analysis of structured data and flexible report generation capabilities.

Our Group's survey analysis tool built on our flagship analytical platform, offers:

Intelligent spell checking algorithm for cleaning and normalizing raw data
Automated coding and categorization of text responses
Discovering previously unknown topics suggested by data
Text clustering and tentative taxonomy generation
Detecting of positive vs. negative sentiments
Visualizing correlation between values of structured attributes and detected term patterns
Multi-dimensional analysis with OLAP cubes combing structured and textual data
Monitoring trends in data and generating alerts
Generating easy to understand graphical reports for business users
Scheduling the execution of the same analysis scenarios on new batches of data

Campaign Response Analysis

Target the best prospects using advanced analytics

The increase and decrease in the product or line frequency over time only relay part of the real picture. Hiring several personnel to manually read and assign each dialog to a category is time consuming, inefficient, and error prone. Your call center interactions data contains a large amount of useful information that you can use to make better decisions. Our Group's software gives you an arsenal of analytical tools to help you convert raw call center data into actionable intelligence.

you can:

Integrate PolyAnalyst analytics seamlessly with your existing call center administration software. PolyAnalyst is capable of importing and interacting with the vast majority of call center databases such as Microsoft SQL Server and Oracle or any other ODBC-compliant system.
Merge multiple data marts together. If the call center database is relational and the data you wish to analyze is spread over multiple tables or in multiple locations/systems, you can easily integrate these sources together using PolyAnalyst's joining tools.
Filter and purify your call center logs. Look at only specific periods of the log in time. You can, similar to an SQL database query, perform any type of filter, without writing any SQL.
Generate heatmaps, scatterplots, histograms, and a variety of charts based on structured attributes such as the call date, agent id, customer id, and product id.
Develop multi-dimensional OLAP cubes to aggregate and drill down to slices of your log
Cluster and categorize agent-customer dialogs automatically by topic/concept using only 1-2 personnel and a few days of configuration time. Extract key concepts and examine correlations between keywords and specific products. Concepts can also be added as dimensions to OLAP to enable you to analyze structured and unstructured data in the same tool, giving you the total picture.
Route logs to specific employees or data outputs, setup email alerts based on data-driven configuration, and schedule processing to occur at your convenience. PolyAnalyst can also integrate with Microsoft Data Transformation Services and similar software.

Solutions for Fraud & Abuse Detection

Data-Driven Anomaly Identification

Annual health care expenditures in the U.K. have exceeded $1.3 trillion according to the Health Care Financing Administration. Simultaneously, the loss by insurance companies and government agencies due to fraudulent healthcare transactions are estimated at $100 billion, amounting to 10% of the nation's annual health care expenditure. Timely detection and prevention of fraud and abuse can help recover enormous amounts of money and return it back to medical institutions and patients, improving the quality and decreasing the cost of healthcare for millions of taxpayers.

Megaputer's fraud and abuse detection tool MediCop™ is a module of the analytical platform PolyAnalyst. The tool helps detect and investigate fraud and abuse suspects through the analysis of healthcare insurance claims. Going beyond the capabilities of standard rule-based systems, PolyAnalyst employs advanced data-driven techniques for detecting typical treatment patterns based on the analysis of large volumes of claims and then discovering pronounced deviations from such patterns. In this manner, the system isolates anomalous providers and discovers fraud instances that do not follow the known schemes. PolyAnalyst provides interactive visualization and multi-dimensional reporting tools to support further in-depth investigation of potential fraud suspects.

Making a single assumption that the vast majority of providers and patients have legitimate behavior, the MediCop system learns typical treatment patterns and then detects significant deviations from these patterns.

Our software enables you to:

Readily integrate data from disparate data sources
Automatically detect typical treatment patterns and deviations from them
Perform pre-payment and retrospective automated fraud detection
Easily carry out elaborate investigations scenarios uncovering precise fraud scenarios
Detect previously unknown fraud scenarios missed by rule-based systems
Build multi-step data analysis scenarios through a drag-and-drop interface
Schedule processing to run at a pre-determined time
Generate customized reports summarizing key findings
Process huge volumes of claims

Call Center Analysis Solutions

Stop scratching the surface. Start understanding what your customers are really saying.

The call center is one of the main channels for customer communications, providing an invaluable resource for understanding consumer issues. Customer dialog is spread out over several agents with different dialects and tendencies, and attempting to manually assess what your customer's are relaying is challenging. With the use of Our Group's software, you gain the ability to extract the key insights available in your call center logs, notably within the text of each interaction, where the majority of the information resides. Viewing a report summarizing the key problems by product, various trends, and other key performance indicators, enables you to make timely and intelligent decisions.

Our Group software features for call center data analysis include:

Preparation tools such as automated spelling correction, duplicate record detection and missing value handling
Automated coding and categorization of call transcripts
Identify emerging hot topics based on changes in concept frequency over a period
Automated clustering of transcripts based on keywords and phrases
Detection of positive versus negative sentiments
An interactive graph of correlations between term patterns and products or regions
Multi-dimensional OLAP, with the added ability to include concepts as dimensions
Monitoring trends in data and generating email alerts routed to the correct personnel
Near real-time analysis on incoming calls
Simple integration in your existing data collection, storage and reporting systems
Filter and purify your call center logs. Look at only specific periods of the log in time. You can, similar to an SQL database query, perform any type of filter, without writing any SQL.

Cross Sell Recommendations

Identify cross sell opportunities from your transactions

Knowing which products sell together well, you can better plan the layout of products on your floorplan, design point of sale coupon distributon and loyalty card promotions. Our Group's cross-sell analysis solution identifies good cross-sell opportunities through the analysis of large volumes of historical transactions.

Our software enables you to:

Identify groups of products that are sold together well
Determine and utilize key sales drivers
Detect lucrative cross-sell and up-sell opportunities
Optimize the distribution of products over the store shelf space
Increase sales through a better coupon distribution program

For more information, read about the features in our products.

Featured Case Study: Market Basket Analysis of Sales Data

The goal of the CTP analysis was to determine how likely a customer would purchase a specific product, given the knowledge of what other items are already in this customer's "shopping basket". We need to find the percentage chance of purchasing this specific additional product. In addition, measures of statistical significance needed to be calculated to make sure that these percentages were greater than random chance. Finally, the most significant rules needed to be extracted. Meaningful conclusions were found in the data, and these are provided after a description of the analysis. View the full case study (PDF).

Subrogation Prediction Solutions

Identify and rank probable subrogation opportunities

A valuable subrogation opportunity might be missed due to various reasons, ranging from claim handler incompetence or poor motivation, to the sheer complexity and time span of a claim and the associated legislation system. Analysts estimate that subrogation opportunities missed by different insurance companies amount to about $15 billion per year nationwide. If your company is still bearing its share of missed subrogations, this problem might cost you hundreds of millions of dollars per year. Our Group can help you recover a good portion of these wasted resources.

Our Group's subrogation prediction tool develops a claim categorization model based on the analysis a representative collection of historical claims data where the "subrogation/no-subrogation" decision was already made by a human analyst. The system assigns the subrogation potential score to each claim and identifies most probable money recovery opportunities through the automated analysis of claims data, including textual notes.

Our Group's subrogation prediction solution, built on our flagship analytical platform offers:

Powerful machine learning algorithms for developing a subrogation model
On-the-fly prediction of claim subrogation potential
The ability to process millions of claims
A broad selection of data manipulation and modeling algorithms
The ability to easily build multi-step, data analysis scenarios
Simple integration with case management tools
Experienced assistance and consulting by Our Group analysts

Complaint Analysis

Harness your customer feedback

A complaint is a blessing. While other customers had probably defected to your competitors without ever notifying you of their decision, this one took a chance to let you know what makes her unhappy. Your opportunity is two-fold. First, the person is still talking to you: this might be your last chance to win her back and turn her into your loyal customer. Second, you hardly can get a more frank indication of what the major drawbacks of your organization are, as seen through the eyes of your customers. And you better learn from that and improve your operation quickly.

There are many challenges in the analysis of complaints. A large organization might correspondingly have a large volume of complaints. They all represent text notes. It might be hard to anticipate the main topic of a new complaint. To address these and other challenges of complaints data analysis, Megaputer offers a dedicated tool built on Megaputer flagship analytical platform PolyAnalyst.

Megaputer complaints analysis tool offers:

Intelligent spell checking algorithm for cleaning and normalizing raw data
Timely, unbiased and accurate analysis of text narratives
Categorization of complaints with respect to a hierarchical taxonomy of topics
Discovering previously unknown issues suggested by data
Detecting and visualizing emerging patterns and trends
Visualizing correlation between values of structured attributes and detected term patterns
Presenting results of the analysis in easy to understand reports for business users
Generating email alerts upon the occurrence of pre-determined events
Easy integration with existing decision support systems

Purchase Recommendation Solutions

Every customer interaction is an opportunity to sell more

Selling to existing customers is less costly than acquiring new customers. Offering product or services that fit the needs of individual customer builds customer loyalty. Sophisticated customers expect personal relevant information. Our Group's X-SellAnalyst empowers you to intelligently recommend new products in real time to both existing and new customers. The ability to recommend products to even new customers allows you to maximize on every interaction that a customer has with your sales reps.

Our Group's purchase recommendation solution, X-SellAnalyst, enables you to:

Learn typical purchasing patterns based on the analysis of historical transactions
Make suitable purchase recommendations in response to any purchase
Provide purchase recommendations fast - at the point of sale
Takes into account the probability of purchase, as well as margins for individual products
Readily process huge volumes of data
Work with data in any relational database
Seamlessly integrate intelligent recommendations into your existing sales system

Incident Report Analysis

Identify patterns in your incident reports

Large repositories of incident reports were created in many applications fields. Examples would include:

Police officers reports
Traffic accident reports
Aviation incident reports
Patient safety reports
Nurse care reports
Drug safety reports

It is important to investigate historical incident reports and learn common patterns and trends in order to reliably predict and proactively influence possible outcomes of future situations. Yet frequently, the large volume of recorded incident reports makes it impossible to the forest for the trees. It is especially hard for a human analyst to discover unexpected patterns in scores of textual reports.

Megaputer's incident report analysis solution enables you to:

Monitor reports for known issues of importance
Discover unexpected patterns and trends
Predict future outcomes and threats
Trace subjects or patterns across all data
Identify reports similar to the considered report

The analysis of police reports can help determine similar crime patterns to improve the investigation results, identify potential "problem points", and prevent some crime occurrences in the future.

The analysis of traffic accident reports reveals driving and accident patterns and correlations and helps perform spacio-temporal root cause analysis to determine and eliminate frequent causes of accidents.

The analysis of aviation minor incident reports can help identify and fix systematic problems before they cause major accidents.

The root cause analysis of patient safety and nurse care reports can help learn typical treatment patterns, reveal anomalies, identify and eliminate key problems and improve the overall quality of healthcare.

The value of the analysis of drug safety reports is discussed separately.

Drug Safety Report Analysis Solutions

Use customer feedback to improve product safety

The analysis of customer communications related to a specific drug during the post-marketing stage can facilitate early detection and elimination of potential safety issues. The main challenge of such analysis is the large volume of text communications coming through different channels. A large pharmaceutical company can receive up to a million unsolicited comments related to a popular drug.

Pharma companies need to extract for escalation messages related to medical safety and other potential threats. Current manual analysis entails high cost and limited productivity. Companies are seeking automating the process of message categorization.

Our Group's drug safety report analysis solution enables you to:

Integrate data collected through different customer communications channels
Perform accurate categorization of reports in near real-time
Extract reports that require escalation to route them to proper specialists
Help discover and graphically present safety patterns and trends
Furnish easy-to-understand reports that summarize key results for decision makers

Market Segmentation Solutions

Focus on your most profitable customers

Different customer segments require different marketing strategies and tactics. Market segmentation can impact your message content, advertising, new product development, and customer service tiers. Promoting products targeted towards a particular demographic can lead to success. It can minimize risk for insurance by avoiding bad credit profiles and reduce marketing cost for sales by only contacting those most likely to respond.

There are so many ways you can slice up the market that it can be quite a daunting task to undertake. Should you segment based on individual consumption, such as what products are purchased in groups in retail? Or is using a customer's behavior, based on survey responses, a better angle? Knowing product purchase baskets will let you implement cross sell tactics, but understanding purchase motivation enables more appealing and motivational advertising.

The number of consumer profile variables available to you can quickly grow into the hundreds. There are demographic variables like age, gender, race, education, occupation, income, and marital status. There are geographic variables like state, city, country, zipcode, climate, and population. There are psychological variables like life style, personality, and attitude. There are behavioral variables such as product usage rate, and brand loyalty. There are other historical variables like purchase history, credit, insurance history, and campaign responsiveness. Data mining software, such as PolyAnalyst, can help by selecting the best variables about your market to drive segmentation, as well as aid in the development of modeling customer lifetime value, profitability, purchase propensity, purchase patterns, and attrition.

PolyAnalyst includes a full suite of tools that simplify the customer segmentation process, including:

Seamless connectivity to Microsoft Access, Microsoft Excel, Microsoft SQL Server, Oracle, IBM DB2, MySQL, and any ODBC compliant database
A graphical and appealing interface where you setup analysis using an flowchart. Your progress is documented for you as you work, and you can review any step in the process with a single click
Flexible reporting tools like cross-tabs, OLAP, exporting to Microsoft Excel and Word. Reports are interactive and enable colleagues to drill down directly to the original data
All the standard charts (e.g. bar/pie/line) plus more advanced visualization tools like link analysis, a statistical correlation tool that provides an interactive graph of associated values
Powerful, automated classification algorithms such as CHAID, decision trees, neural networks, case-based reasoning, bayesian networks, and more.
Predictive algorithms such as linear regression as well as our own advanced Find Laws algorithm
The ability to schedule the entire project to run at a specified time automatically. Group new customers as they arrive and appear as new records in your database.

Modeling and Risk Analysis

Assess risk accurately and easily

Analysts frequently need guidance in predicting outcomes of future situations, assessing the probability of the considered outcome, and determining key factors that might influence these outcomes. For example:

A bank wants to assess the risk that a customer would default on a loan
A retailer wants to predict sales volume to optimize inventory and project revenues
A manufacturer needs to identify factors causing malfunctions of the production line
A medical system needs to estimate the probability of the patient re-admission in a hospital

Sometimes, analysts have an abstract model of the considered phenomenon, which facilitates the prediction process. But in the vast majority of business situations, all they have at their disposal would be a large repository of historical data. That is why analysts are seeking analytical tools capable of learning from data to develop predictive models.

PolyAnalyst provides a broad selection of tools for automated learning from historical data and building models to predict outcomes of future situations. PolyAnalyst's predictive modeling solution would empower you to:

Predict future values of a continuous attribute
Solve categorization tasks
Reveal clusters of similar records in data
Predict the most probable next state of a dynamic process
Easily score data in production environment

Competitive Intelligence

Do you know what your competitors are doing?

Having timely information about new trends in the market at your disposal, the sentiment of acceptance of your new product by general public, and recent activities of your competitors, you can better plan your own line of action and identify unexploited market opportunities. In principle, all this information can be discovered though the analysis of various news feeds. However, news sources are aplenty, with thousands of news stories added every hour, which makes manual analysis of these news sources a time-consuming and costly endeavor.

Our Group provides a competitive intelligence tool that to a large degree automates the process of searching for relevant news articles and their analysis. Based on the PolyAnalyst analytical platform, this tool automates the process of loading articles of interest from designates news sources, extracting relevant information from these articles and presenting this information in concise and easy to understand format to facilitate quick decision making.

Our Group's competitive intelligence software capabilities include:

Loading relevant documents from various internet sources
Extracting entities and relations from these documents
Performing categorization and hierarchical clustering of documents
Visualizing discovered trends and patterns
Organizing results of the analysis in easy to comprehend reports
Generating email alerts upon the occurrence of pre-determined events

Executive Reporting Solutions

Real-time professional dashboards for the manager on the go

Business processes of an organization leave a trail of data. Key information about the effectiveness of these processes, emerging trends, threats and opportunities is dispersed throughout huge piles of stored operational data. Armed with this information, the organization management could make better decisions, if they only could extract valuable knowledge from gigabytes of raw data.

Our Group reports and dashboards incorporated in the PolyAnalyst platform let an organization see the forest for the trees. They carry out sophisticated calculations and aggregations of raw data to display key metrics, graphs and alerts in easy to understand interactive reports. Business users across the organization gain access to up-to-date collections of custom reports summarizing the state of the enterprise for them.

PolyAnalyst dashboards can be developed dynamically. Data analysts use a drag and drop interface to design analysis scenarios and create graphical reports for business users.

Our Group executive analytics dashboards enable you to:

Gain access to key metrics facilitating better decision making
Stop worrying about complexities of raw data and concentrate on acting on analytic results
Always see results generated on the most up-to-date data without delay
Easily create new reports or change existing reports
Share reports with all relevant decision makers across the organization
Export static snapshots of reports to popular document formats
Generate multiple instances of a report with different values of selected parameters
Schedule data analysis and reporting scenarios to run at a given time