The Department of Statistics at Iowa State University will host the Conference on Predictive Inference and Its Applications on May 7 and 8, 2018, in Ames, Iowa. The conference program includes plenary presentations from 16 distinguished speakers and a poster session featuring presentations from students, postdoctoral scholars, and early-career researchers. Goals of the conference include raising awareness about the importance of prediction, showcasing research of current and emerging leaders in the field, motivating the development of more accurate prediction methods, and encouraging interactions and collaborations among a diverse collection of scientists with complementary skills and abilities. In addition to attracting PhD statisticians and graduate students majoring in statistics, the conference has the potential to draw participation from the broader data science community that includes researchers from bioinformatics, business analytics, computer science, electrical and computer engineering, economics, and social sciences among other areas. The Conference on Predictive Inference and Its Applications will provide a valuable venue for developing new methodologies that enable accurate prediction and assessment of uncertainty in our data-rich world. Such methodologies provide valuable insights and lead to better decision making in science and industry.
Problems involving the prediction of unobserved but eventually observable quantities are ubiquitous in the modern world. Historically, the discipline of statistics has placed greater emphasis on hypothesis testing and parameter estimation than it has on prediction. Other research communities connected to but not equivalent to statistics, including computer science, machine learning, analytics, big data, data mining, and data science have more fully embraced prediction problems and developed a variety of useful prediction tools. Nonetheless, statistical thinking and innovation have an important role to play in addressing prediction problems now and in the future. In some cases, statistical thinking can be used to generate appropriate measures of uncertainty to accompany point predictions provided by existing tools. In other cases, statistical thinking can lead to methods for predictive inference that strike a better balance between bias and variance or make more efficient use of the data than existing approaches. The formal presentations and ideas that arise from discussions at the conference are intended to motivate scientific progress and result in publications that will appear in refereed journal articles. This research will enhance the state of the art in prediction methodology, which in turn will lead to advances in many fields where accurate predictions and accurate assessments of prediction error are essential. The conference website is available at PredictiveInference.github.io.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
