Twelfth Industrial Mathematical and Statistical Modeling Workshop for Graduate Students

The volume of traffic on a particular website often follows some pattern such as Time of the Day (TOD) or Day of the Week (DOW) trends. We developed three mathematical models to capture these patterns, so we can predict future volume based on previous data. The first model is based on Fourier transformations [7],[8], andFig. it shows good performance when the main frequency of the wave is relatively low. The second model is based on velocity and can adjust the prediction with the latest available data. The third model uses polynomial fitting [5] to predict DOW behavior and will also adjust the predictions according to the latest data. Louisiana State University Georgetown University Ohio University Auburn University Purdue University University of Louisville