Project Showcase

For a major bank, we used Business Analytics to more precisely identify a target market for a campaign that increased sales by 200%. By analyzing data on customer preferences, such as purchasing data from points of sale systems and credit card transactions, we predicted who is most likely to respond to a particular marketing strategy. For this project, we applied:

	Classification - Decision Tree Models
	Cluster Analysis

We enabled a consumer products company to increase cross selling of two lines by 100%. We cross referenced products to reveal consumer buying patterns. We showed how certain products sell together and that knowledge led to increased sales.  For this project, we applied:

Association Models

We substantially decreased defects for an automotive parts manufacturer by applying analytics on historical production data. Our analysis of attributes revealed key parameters that contributed to quality of goods produced. By controlling the key parameters we identified, the manufacturer cut defects substantially. For this project, we applied :

Stepwise Regression Analysis

For a state Department of Transportation, we developed a comprehensive performance management system that continuously improves performance at all levels, helps DOT deliver cost effective services and projects, improve public confidence and responsiveness. 

The underlying cost analysis enabled the DOT to understand true costs and that improved the use of funds.

	Activity-based costing model
	Process improvement
	Strategy map & cascaded scorecards
	Analytics & forecasting
	Integrated reporting functionality including GIS
	Upfront, quantified ROI
	Sensitivity to existing culture & change management

SAS Activity-Based Management

We worked with a pharmaceutical company to analyze the quality of products and services using data such as repair records, customer complaints, and the number of returns. Problem products or areas were then studied in detail for improvement. The results reduced wastage by 60% and saved hundreds of thousands of dollars.  For this project, we applied:

	Classification-Decision Tree Models
	Regression

For a Clinical Trials Company, we analyzed historical data from different sources  to build a forecasting model for future patient enrollment, demand of downstream and upstream materials. We forecasted weekly material requirements for the coming year.  The results improved inventory management of dated materials leading to 40% decrease of unutilized inventory.  For this project, we applied:

	Trend, decomposition and correlation analyses
	ARIMA
	Growth fitting curves
	Regression analysis (linear and non-linear)
	Confidence limits
	Neural Network
	Bayesian probabilities and survival rates
	Periodic parameter optimization
	Model back-testing

An oil company commissioned us to build a system to manage sales in India. It worked so well, the system was extended to another company... and then another and then across another continent until the same solution was managing sales on five continents for the company. Headquarters could drill down to timely details in the far reaches of its sales. 

	Found the most profitable allocation of marketing funds
	Arrived at an optimal price point to maximize total profits for products
	Forecast the effect of different promotional activities on future sales
	Analyzed daily point of sales data to project aging stock and potential stock-out situation

We analyzed for an insurance company historical data of different modules to quantify the various sources of risk within a group and project its risks into the future.  We discovered that in a worst case scenario, loss could be 80% more than the average estimated. By identifying the actual risk, we allowed the insurance company to substantially decrease its exposure to risk. For this project, we applied:

	Monte Carlo simulation of the underlying exposures in several key areas
	Risks simulated independently using relevant distributions (normal, lognormal, etc.)
	The rank correlation method involving the Cholesky decomposition used to correlate risks

For a large financial services company, we analyzed the historical data of CDOs performance and deal structures to forecast the performance of different tranches. We helped the company to optimize an investment portfolio according to an investor’s profile.  We used prices of different tranches and their expected performances along with their expected loss distributions  to determine the optimal tranches for investment given the investor’s risk aversion profile.  For this project, we applied:

	Optimization techniques
	Bayesian probability analysis
	Scenario analysis and loss distribution using waterfall and scenario building methods based on several factors including   coupon rates and spreads