Massachusetts Analytics developed a modeling system to forecast market-wide and company-by-company sales opportunities for a variety of financial services products.

Challenge

A consulting company provides its financial services clients with a business intelligence tool providing strategic and account level analytics.  The company had access to valuable data detailing firm-by-firm usage of various financial services products.  They needed to develop an analytics package for the business intelligence tool that used this data to provide insights into market and firm level opportunities.

Modeling challenges included:

• Interactions between independent variables
• Non-linear relationships between dependent and independent variables
• Inability to observe the dependent variable directly

Solution

Massachusetts Analytics created a modeling system using a combination of linear and non-linear regression, including non-parametric regression techniques to develop forecasts of product usage for companies based on various company specific features.

Data issues were addressed with the creation of various sub-models estimating the dependent variable based on a variety of features.

The model included the flexibility to incorporate the company’s substantial market expertise and qualitative estimates into the model forecasts.

Value Created