Page 276 - Data Science Algorithms in a Week

Agent-Based Modeling Simulation and Its Application to Ecommerce 257

The complexity of the market and customer behaviors benefit from nontraditional
modeling tools for analysis. Behaviors can be defined at individual level and at the
system level. Hybrid simulation provides an approach that does not make the assumption
of a perfect market and homogeneity.
Internet based models cause disruptions to traditional business models. New players
find it challenging navigating the highly competitive landscape of this complex
environment. Due to aforementioned characteristics, the ecommerce system tends
towards complexity. There exist several performance risks associated with the business
model. These risks include minimal return on investment, government regulations and
lack of trust. Results from case studies and literature review reveal that the performance
of C2C ecommerce remain under explored from a system perspective. Complex
interactions exist among stakeholders, the changing environment and available
technology. There is a need for an integrated system that will provide a testing ground for
managing control actions, anticipating changes before they occur and evaluating the
effects of user actions on the system at different managerial levels.
The presence of continuous and discrete behaviors poses challenges for the use of
existing simulation tools in simulating the C2C ecommerce space. The system is
characterized by uncertainty as well as government regulations and external factors.
Important factors such as liquidity and different threshold values for consumers remain
undefined. Not addressing these issues can result in financial losses and lack of trust that
can erode the benefits of the business model. There is a need to systematically map,
model and evaluate the viability and performance in order to realize the best tradeoff
between benefits and risks. This study presents a framework to systematically map,
model and evaluate the viability and performance in order to evaluate tradeoffs between
benefits and risks.
The paper is organized as follows. Section 2 introduces the application of system
simulation and modeling (system dynamics in particular) to ecommerce research. Section
3 describes the developed framework. Section 4 presents the Lending Club case study
while the application of the agent based simulation and the system dynamics models as
well as some results are presented in Section 5. The paper concludes and prescribes some
future directions for this study in Section 6.

BACKGROUND

Classifications of System Simulation and Modeling

Ecommerce systems are intelligent systems (Bucki and Suchanek, 2012) and a
system (simulation) can be discrete or continuous. In continuous simulation, the system
evolves as a continuous function represented by differential equations while in discrete