Staffing Call Center Agents

I took a class on operations research models, and one of the areas we covered was stochastic modeling. We looked at calls arriving in a call center as one example where this could be applied and used to optimize resources.

A team of three mechanical engineering undergraduates and I obtained and cleaned a data set of calls from an Israeli bank, with the goal of identifying the optimum number of bank representatives that should be staffed at any given hour to minimize the time that a customer had to wait to resolve their issue. At the same time, we wanted to minimize this number so as to avoid too high of a cost to the bank.

The bank received calls throughout the day, and calls were labeled by type — some were for stock activity, others were informational inquiry, etc. We modeled call arrival and processing as discrete-time Markov chains, and considered how the call center would fare if customer service representatives were specialized to only handle one type of call, or able to answer all types of questions.

We created and delivered a presentation summarizing our results, which you can see below: