Bayesian Hierarchical Models for Predictive Analysis

Abstract: Many predictive models required to analysis data that are structured in groups and clusters. In such a data structure each group has it is own parameters however the parameters are related because of hierarchical structure of data. In this case two modeling approached might be possible:

To model all data together
To model each group separately.

The concern with the first approach is ignoring both autocorrelation and latent differences while the second approach the similarity is ignored and data sparsity can be problematic. The best approach for such a situation is developing a hierarchical model which has the flexibility to capture and analyze this data structure and the ability to account for and estimate effects from different groups. Hierarchical models borrow strength across the group and as the result minimize the effect of data sparsity. Hence the information of higher level can be share effectively among the lower level group while lower level estimation still follows their own structure and pattern.

In this session we focus on the application of Bayesian hierarchical linear regression model in the area of pricing and revenue management. We will discuss how multi-layer model can be applied to hierarchal dataset to deal with data sparsity and reduce the noise to provide reliable and robust prediction. In this use case the behavior of every individual is modeled as a linear regression assuming each individual has its own unique behavior while there are some similarities in the behavior of the same groups.

Bio: Amir Meimand is Zilliant Director of R&D, pricing scientist, where he designs and develops pricing solutions for customers and performs research in which he applies new methods to improve the current solutions as well as develop new tools. Prior to joining Zilliant, Amir helped design and develop a promotion planning and pricing platform for B2C retailers.

Amir holds a dual Ph.D. degree in Industrial Engineering and Operations Research from Pennsylvania State University. In his doctoral work, he applied operations research concepts to dynamic pricing and revenue management.

Open Data Science Conference