Lessons learned from using Tweedie’s compound Poisson model

Event Date: 
Friday, 28 February, 2020 - 12:00
Location: 
IBMI
Lecturer: 
Assist Prof Michael D. Burnard, InnoRenew CoE and University of Primorska, Institute Andrej Marušič

Data is rarely as straightforward as we hope or as we are usually taught. One of the biggest challenges I’ve faced in my academic career was analyzing the data from my Master’s research. Without this challenge, I’m not sure I would have developed such an interest in statistics or data science. My Master’s research addressed a problem that may seem simple on the surface: how do we quantify a certain type of defect in the surface of wood furniture? My task was to develop a method to measure a particular phenomenon, but my advisors hadn’t envisioned any particular challenges in analyzing the data once it was collected. Apart from the data collection challenges (which will be mentioned briefly), there were several challenges with the data. First, we weren’t sure which attribute to include in our model the number of defects detected or the size of the defects (does one matter more to customers?) – we had both count data and continuous data. Second, there were a lot of zeroes in our data – the defect didn’t always occur. Third, our experiment had mixed effects – not all features were evenly distributed across all specimens (we used a blocked split-plot design). Finally, our data was not parametric, making determining both point estimates and confidence intervals challenging – the eventual solution was bootstrapping. This presentation will introduce the real-world problem (cracks in decorative veneered plywood) along with the data collection methodology (digital image correlation), the experimental design challenges, and the analytical approach taken.

Related outputs:
Burnard, M.D., Muszyński, L., Leavengood, S., Ganio, L., Burnard, M.D., (2018). An optical method for rapid examination of check development in decorative plywood panels. Eur. J. Wood Wood Prod. 0, 0. https://doi.org/10.1007/s00107-018-1327-7
Burnard, M., Leavengood, S., Muszyński, L., Ganio, L., (2019). Investigating face veneer check development in decorative plywood panels: the impact of four common manufacturing factors. Eur. J. Wood Wood Prod. 77, 961–979. https://doi.org/10.1007/s00107-019-01455-2
Burnard, M. & Ganio, L. (2019) Inspecting, Plotting, & Modelling Check Density in Decorative Maple Veneered Plywood Panels. [Online]. https://doi.org/10.6084/m9.figshare.6964862.v1

 

About IBMI

Institute for Biostatistics and Medical Informatics (IBMI), formerly Institute for BioMedical Informatics (so still IBMI) was founded by the Faculty of Medicine as a result of a need for a unit which would perform, or coordinate, tasks related to data analysis and providing information, relevant for research in medicine. The programme of the institute, and its development, have been adjusting thorugh time to changes in financing and technological progress, but the basic aim remain the same: to support research in medicine. This is achieved through the following tasks:

Contact

Institute for Biostatistics and Medical Informatics
University of Ljubljana, Faculty of Medicine
Vrazov trg 2, 1000 Ljubljana
Slovenia

tel: +386 1 543-77-70
fax: +386 1 543-77-71
email: ibmi (at) mf.uni-lj.si