scholarly journals Knowledge in Imperfect Data

10.5772/37714 ◽  
2012 ◽  
Author(s):  
Andrzej Kochanski ◽  
Marcin Perzyk ◽  
Marta Klebczyk
Keyword(s):  
Imperfect Data

The effect of imperfect data on default prediction validation tests

2012 ◽  
Vol 6 (1) ◽  
pp. 77-96 ◽  
Author(s):  
Heather Russell ◽  
Douglas Dwyer ◽  
Qing Kang Tang
Keyword(s):  
Imperfect Data ◽  
Default Prediction ◽  
Validation Tests

scholarly journals Consideration of uncertainties in environmental science and management with examples from Pacific salmon

2015 ◽  
Author(s):  
Randall Peterman
Keyword(s):  
Quantitative Methods ◽  
Environmental Science ◽  
Pacific Salmon ◽  
Natural Populations ◽  
Observation Error ◽  
Imperfect Data ◽  
State Dependent ◽  
Strategy Evaluation ◽  
Management Procedure ◽  
Management Procedures

An informal review of the history of new quantitative methods in environmental science, including environmental risk assessment, shows about a 10- to 20-year lag in wide acceptance of such methods by management agencies. To reduce that lag time as innovative methods continue to emerge, environmental scientists will need to work much more intensively with communications specialists on better ways to explain risk analyses and decision-making strategies to non-technical decision makers and the public. Four key uncertainties make such communication difficult: (1) natural variability in both physical and biological processes, (2) imperfect data arising from observation error (i.e., measurement error), (3) incomplete understanding of an environmental system's structure and dynamics, and (4) outcome uncertainty (deviations between realized outcomes and management targets). These uncertainties create risks -- risks to natural populations as well as to people who use them. Examples of these four sources of uncertainty are presented here for Pacific salmon (Oncorhynchus spp.). One promising framework for explicitly taking such uncertainties into account was initially developed in the early 1990s by scientific advisors to the International Whaling Commission. They built stochastic models, which essentially were comprehensive formal decision analyses, to derive management procedures (i.e., sampling designs for collecting data, methods to analyze those data, and state-dependent harvest-control rules for use by managers) that were robust to all the uncertainties considered. This method of "Management Strategy Evaluation" or "Management Procedure Evaluation" is now considered the "gold standard" for conducting risk assessments and making risk-management decisions in marine fisheries.

Exploring incomplete information in maintenance materials inventory optimization

2014 ◽  
Vol 114 (1) ◽  
pp. 144-158 ◽  
Author(s):  
Antti Puurunen ◽  
Jukka Majava ◽  
Pekka Kess
Keyword(s):  
Sensitivity Analysis ◽  
Simulation Model ◽  
Input Data ◽  
Real Life ◽  
Significant Risk ◽  
Service Level ◽  
Content Type ◽  
Imperfect Data ◽  
Inventory Optimization ◽  
Critical Materials

Purpose – Ensuring the sufficient service level is essential for critical materials in industrial maintenance. This study aims to evaluate the use of statistically imperfect data in a stochastic simulation-based inventory optimization where items' failure characteristics are derived from historical consumption data, which represents a real-life situation in the implementation of such an optimization model. Design/methodology/approach – The risks of undesired shortages were evaluated through a service-level sensitivity analysis. The service levels were simulated within the error of margin of the key input variables by using StockOptim optimization software and real data from a Finnish steel mill. A random sample of 100 inventory items was selected. Findings – Service-level sensitivity is item specific, but, for many items, statistical imprecision in the input data causes significant uncertainty in the service level. On the other hand, some items seem to be more resistant to variations in the input data than others. Research limitations/implications – The case approach, with one simulation model, limits the generalization of the results. The possibility that the simulation model is not totally realistic exists, due to the model's normality assumptions. Practical implications – Margin of error in input data estimation causes a significant risk of not achieving the required service level. It is proposed that managers work to improve the preciseness of the data, while the sensitivity analysis against statistical uncertainty, and a correction mechanism if necessary, should be integrated into optimization models. Originality/value – The output limitations in the optimization, i.e. service level, are typically stated precisely, but the capabilities of the input data have not been addressed adequately. This study provides valuable insights into ensuring the availability of critical materials.

Changing Spectrum of Serious Child Abuse

PEDIATRICS ◽  
1986 ◽  
Vol 78 (4) ◽  
pp. 715-716
Author(s):  
ABRAHAM B. BERGMAN
Keyword(s):  
Child Abuse ◽  
Physical Abuse ◽  
Data System ◽  
Medical Examiner ◽  
King County ◽  
Imperfect Data ◽  
Clinical Problems

In Reply.— Dr Hanson seems to be so upset with our article that she appears not to have read it. I thought we packed our paper with disclaimers of our imperfect data system and our fallibility in identifying perpetrators. In epidem research on clinical problems, one must play with the cards that are dealt. The diagnosis of physical abuse is a medical one and was made by the pediatrician attached to the child abuse teams or by the King County medical examiner.

Effective Intelligent Data Mining Using Dempster-Shafer Theory

2008 ◽  
pp. 2943-2963
Author(s):  
Malcolm J. Beynon
Keyword(s):  
Data Mining ◽  
Missing Values ◽  
Uncertain Reasoning ◽  
Worst Case ◽  
Data Mining Techniques ◽  
Imperfect Data ◽  
Dempster Shafer Theory ◽  
External Management ◽  
Shafer Theory

The efficacy of data mining lies in its ability to identify relationships amongst data. This chapter investigates that constraining this efficacy is the quality of the data analysed, including whether the data is imprecise or in the worst case incomplete. Through the description of Dempster-Shafer theory (DST), a general methodology based on uncertain reasoning, it argues that traditional data mining techniques are not structured to handle such imperfect data, instead requiring the external management of missing values, and so forth. One DST based technique is classification and ranking belief simplex (CaRBS), which allows intelligent data mining through the acceptance of missing values in the data analysed, considering them a factor of ignorance, and not requiring their external management. Results presented here, using CaRBS and a number of simplex plots, show the effect of managing and not managing of imperfect data.

Sign in / Sign up

Export Citation Format

Share Document