Checking Data Informativity as the First Step in Data-Driven Modeling – Case Study

EasyChair Preprint 8885

Abstract

This paper reviews the strategies for testing a given dataset sampled from an unknown dynamic process to determine if it is sufficiently informative to model the system’s behavior. The presented test should be done as the first step in data-driven modeling to avoid an endless search for a proper model which may not exist based on available data. Furthermore, since measuring all process variables to form a complete set of information for modeling purposes is impossible in many cases, these tests can also estimate what someone can expect from the model established based on given data. Finally, the presented methodologies are applied to a real process as the case study.

Keyphrases: data informativity, data-driven modeling, feature selection, industrial process, system identification, timestamp dataset

@booklet{EasyChair:8885,
  author    = {Amir Farzin and Kateryna Rabchuk and Bernt Lie and Nils-Olav Skeie},
  title     = {Checking Data Informativity as the First Step in Data-Driven Modeling – Case Study},
  howpublished = {EasyChair Preprint 8885},
  year      = {EasyChair, 2022}}