Past Projects

Details and references and client list available on request for projects outside of any Non-disclosure Agreement.

List contains only distinct project and not those that were repeated.



  • Attrition and Revenue forecasting for retail bank – produce stable monthly forecast that uses not only time-serious components but also variety of internal and external causals.
  • New product demand and sales forecasting – produce new business forecasts for different banking products and the different levels of organizational hierarchies.
  • Text Mining Banking Surveys – Text Mining research survey to establish connection between different words, phrases, entities and low or high satisfaction customer level – on different organizational level. Furthermore, using text mining technologies to segment customer feedback based on comments given.
  • Anger Mining Social Media – search for extreme adverse words, phrases, entities and linking them to products or services at specific level of organizational structure.
  • Fraud detection – Building custom fraud models in online and credit card areas to build on and improve on existing fraud defense systems.
  • Human resources – Assisting in building models that would predict resignations or promotions  and extract meaningful patterns that would lead to these.
  • Text Mining for Customer Insights – Creating variety of descriptive models that could lead to better understanding of what is causing customer to leave for competitors.
  • Customer segmentation and cross-sell – produce different customer segments and overlay it with appropriate product and look-alike models to boost existing cross-sell efforts.
  • Anomaly detection – use multiple analytical approaches to detect anomalies for purposes of improving on fraud-detection.
  • Credit Risk – building and assisting in building different types of credit risk models (PD’s, LGD’s, CCF)


Image result for investment banking

  • Early warning financial distress – produce models that will timely identify clients who are entering financial difficulties based on external data and internal financial indicators
  • Lead generation model – use data science to differentiate between client who took up specific financial product and those who did not and use it for prospect ranking and lead generation
  • Equity capital raise model– predict which companies are likely to raise specific type of equity capital.



INSURANCE (Life / Short Term / Healthcare):


  • Risk Rating – supplement actuarial team with risk rating models for pure premium, loss-ration, claim frequency and severity
  • Policy Lapse Prediction Modeling – build models that would predict policy lapse within specific period of time.
  • Claim Segmentation – automatically segment suspicious, high value and ordinary claims before submitting it specific types of digital or mobile adjudication.
  • Analysis of driving behavior - analyzing driving sensor data to uncover the patterns that lead to accidents for insurance purposes
  • Fraudulent Claims identification and scoring– creating predictive model that can identify potential set of fraudulent claims.
  • Intermediary/Broker performance analysis - profile brokers based on its past performances using set of input variables.
  • Dormancy Reactivation – Create two-stage model that can identify segment of the customer within dormant population that can still be reactivated with appropriate marketing stimuli.
  • Client Potential – Create models that can assign future buying potential to specific customer segment.
  • Response modeling – develop champion response model that can further increase current response rate.
  • Competitive Intelligence – text-mining on competitive data from online sources.
  • Death Prediction - predict death of insured within specific customer segment and estimate the cost of death.
  • Hospital allocation – optimize costs and resources by allocating specific medical procedures to number hospitals based on set of criteria.
  • Clinical and Financial Outcome analysis - analyze specific set of factors that lead to specific outcomes (desirable vs. undesirable)
  • Disease Profiling - analyzing set of factors that lead better management of specific disease and faster recovery
  • Medical Error analysis - analyzing clinical patterns that lead to preventable adverse effects



  • Customer churn (pre paid segment) - creating champion model that can predict customer churn within period of time given the set of enritched input variables.
  • Recharge prediction - predict customer's amount of recharge within period of time.
  • Text-mining on social media and call centre data - segment social media comments as well as call centre data based on some underlying actionable topic such as account issues, network coverage, etc.
  • Revenue and Attrition Forecasting - forecast revenue and attrition on different hierarchical and geographic levels.
  • Customer segmentation - segment customer for new product development or marketing offers.



  • Customer Behavior Modeling and Market Basket Analysis
    Aim of the project was to produce natural customer segments and to better understand customer behavior, product portfolios and product associations.
  • Retail store segmentation – Objective was to segment stores based on product demand as well as on demographic and other characteristics of the store catchment area.
  • Segmentation migration – Produce the set of models that would predict which of the members of the existing value-based segments are likely to be migrated in higher value segments. Also, produce additional models that would identify segment members that are likely drop off in lower value segments. And lastly profile the segment's core.
  • Demand Forecasting – Produce the set of forecasting models that would improve store demand planning, so that problems stock-outs and over-stocking is reduced.
  • Analytics for E-tailer - Build the variety of models that would increase revenue and improve sales, as well help with recommendations, product categories, and improve overall customer-centricity of the organization.
  • Sentiment Analysis for E-tailer – Text Mine online customer segments for specific retailer and for his main competitors and present actionable findings and insights.
  • Internal Customer Acquisition, Look-alike-modeling and Cross-sell – Retail organization has opened new line of products. Objective was to build set of models that would produce look-alike score between small base for the new line of business and rest of the customer base. Customers within rest of the base that have look alikeness score above specific threshold would be good candidates for promoting new line of business.
  • Market basket analysis for clothing retailer – identify actionable, and non-trivial association between different items that are outside of promotional cycle.
  • Forecast short-term demand in replenishment area to improve on supply-chain activities
  • Produce marketing and behavioural segments for card customers and analyze their purchasing patterns, and do RFM modeling and market basket analysis.



  • AIRLINE: Building different set of models for airline: Late On Gate model, Early Birds Model, Route Profitability models, Forecasts of Pax/RPK per class, route. Profiling profitable vs unprofitable route, Load Factor prediction per routes and classes.
  • LENDERS: Building of  Credit Application models, Behaviour models, Debt Collection, Credit response - build models that would predict policy lapse within specific period of time.
  • GOVERNMENT: Building different types of models to assist in various aspects of service delivery for different government department.
  • HOSPITALITY: Produce customer insights, segments and personas for leading hotel group and do extensive patron analysis using booking and loyalty information.
  • MINING INDUSTRY - create early warning system fthat will generate alert of the looming defect or breakdown. Telephony analysis to detect fraudulent rings.
  • GOVERNMENT / DEFENSE - Use external data to identify patterns that would lead to war/unrest.
  • ANALYSIS OF CHILDREN'S HOSPITAL DATABASE - extract patterns that lead to different categories of child related injuries that could be used in injury prevention programs.
  • CUSTOMER ANALYSIS FOR CINEMA CHAIN - viewership analysis for different cinemas based on movie characteristics based.
  • ENERGY SECTOR - forecast demand for electricity consumption, profile out-of-norm users (distribution), predict unplanned outages on generating level unit. Customer sentiment analysis of social media data.
  • ANALYTICS FOR CASINOS - patron analysis, segmentation and profiling.
  • MODELING FOR WELLNESS INDUSTRY - Employee pattern analysis which shows data patter associative to the specific wellness issue (depression, sickness, alcohol-related problems, absenteeism, etc)
  • AUTOMOTIVE - text mine car service data and identify issues with specific repair outlets for specific car models, do customer segmentation, purchasing patterns and affinity analysis using demographic, lifestyle and product data
  • PUBLISHING - forecast demand for specific magazine for specific outlet and optimize the model
  • LAW ENFORCEMENT- crime analysis using time and spatial information together with victim and perpetrators characteristics to unearth actionable characteristics for crime prevention purposes. Build the model to predict ATM robberies.