Facebook Instagram Twitter RSS Feed PodBean Back to top on side

Aktuality

Ilustračná snímka

Pozvánka na prednášku prof. Lynna Roya LaMotta

29. 11. 2022 | zhliadnuté 156-krát

Ústav merania SAV, v. v. i., pozýva na prednášku Inverse prediction from multivariate, heteroscedastic responses, ktorú bude mať profesor Lynn Roy LaMotte (USA) vo štvrtok 1. decembra 2022 o 10.00 hod., v zasadacej miestnosti Ústavu merania SAV.


Profesor Lynn Roy LaMotte je emeritný profesor programu bioštatistika na Louisianskej štátnej univerzite (LSU), Health Sciences Center, School of Public Health, v New Orleans, Louisiana, USA. Medzi jeho hlavné výskumné aktivity patrí štatistická analýza, štatistické modelovanie, aplikovaná štatistika, analýza údajov, matematická štatistika, viacrozmerná štatistika, analýza viacrozmerných údajov, lineárna regresia, štatistická inferencia, logistická regresia, regresná analýza, zmiešané lineárne modely, lineárne odhady, kategorická odpoveď, forenzná štatistika.

Inverse prediction from multivariate, heteroscedastic responses

(abstract)

Inverse prediction is a method for inferring, from a subject’s response, what population that subject came from. To be useful, such inference must include assessment of variability, like a confidence set or a p-value table or graph.

Example: based on training data from insect larvae of known ages, and given measurements from a single mystery specimen, estimate the age of the mystery specimen.
Example: based on measurements of biparietal diameter and femur length from 1100 ultrasound scans of fetuses in vivo of known gestational ages, and given the measurements from an USS of a fetus of unknown GA, guess its GA.
Example: given measurements of percent DNA methylation at two genetic loci from saliva samples of 91 persons of known ages, guess the age of a mystery specimen.

Models for such processes typically need to be flexible, multivariate, and allow for variance-covariance matrices that change with age. I’ll describe the statistical setting, how that can be managed in widely-available statistical software, like SAS, BMDP, Stata, SPSS, and R, and give graphical examples.

 

Spracovali: Beáta Ondrušová a Lukáš Zelieska, UM SAV, v. v. i.

Súvisiace články