Nuclear forensics analysis with missing data

Roisin T. Langan; Richard K. Archibald; Vincent E. Lamberti

doi:10.1007/s10967-015-4458-x

Nuclear forensics analysis with missing data

Roisin T. Langan, Richard K. Archibald, Vincent E. Lamberti

Data Analysis and Machine Learning

Research output: Contribution to journal › Article › peer-review

2 Scopus citations

Abstract

We have applied a new imputation-based method for analyzing incomplete data, called Monte Carlo Bayesian Database Generation (MCBDG), to the spent fuel isotopic composition (SFCOMPO) database. About 60 % of the entries in SFCOMPO are absent. The method estimates missing values of a property from a probability distribution created from the existing data for the property, and then generates multiple instances of the completed database for training a machine learning algorithm. Uncertainty in the data is represented by an empirical or an assumed error distribution. The method makes few assumptions about the underlying data, and it compares favorably against results obtained by replacing missing information with constant values.

Original language	English
Pages (from-to)	687-692
Number of pages	6
Journal	Journal of Radioanalytical and Nuclear Chemistry
Volume	308
Issue number	2
DOIs	https://doi.org/10.1007/s10967-015-4458-x
State	Published - May 1 2016

Keywords

Bayesian methods
Machine learning
Missing data
Monte Carlo methods
Nuclear forensics
Spent fuel isotopic composition (SFCOMPO) database

Access to Document

10.1007/s10967-015-4458-x

Cite this

@article{60952ecd066348b692b8cbdd79293882,

title = "Nuclear forensics analysis with missing data",

abstract = "We have applied a new imputation-based method for analyzing incomplete data, called Monte Carlo Bayesian Database Generation (MCBDG), to the spent fuel isotopic composition (SFCOMPO) database. About 60 % of the entries in SFCOMPO are absent. The method estimates missing values of a property from a probability distribution created from the existing data for the property, and then generates multiple instances of the completed database for training a machine learning algorithm. Uncertainty in the data is represented by an empirical or an assumed error distribution. The method makes few assumptions about the underlying data, and it compares favorably against results obtained by replacing missing information with constant values.",

keywords = "Bayesian methods, Machine learning, Missing data, Monte Carlo methods, Nuclear forensics, Spent fuel isotopic composition (SFCOMPO) database",

author = "Langan, {Roisin T.} and Archibald, {Richard K.} and Lamberti, {Vincent E.}",

note = "Publisher Copyright: {\textcopyright} 2015, Akad{\'e}miai Kiad{\'o}, Budapest, Hungary.",

year = "2016",

month = may,

day = "1",

doi = "10.1007/s10967-015-4458-x",

language = "English",

volume = "308",

pages = "687--692",

journal = "Journal of Radioanalytical and Nuclear Chemistry",

issn = "0236-5731",

publisher = "Akad{\'e}miai Kiad{\'o}",

number = "2",

}

TY - JOUR

T1 - Nuclear forensics analysis with missing data

AU - Langan, Roisin T.

AU - Archibald, Richard K.

AU - Lamberti, Vincent E.

PY - 2016/5/1

Y1 - 2016/5/1

N2 - We have applied a new imputation-based method for analyzing incomplete data, called Monte Carlo Bayesian Database Generation (MCBDG), to the spent fuel isotopic composition (SFCOMPO) database. About 60 % of the entries in SFCOMPO are absent. The method estimates missing values of a property from a probability distribution created from the existing data for the property, and then generates multiple instances of the completed database for training a machine learning algorithm. Uncertainty in the data is represented by an empirical or an assumed error distribution. The method makes few assumptions about the underlying data, and it compares favorably against results obtained by replacing missing information with constant values.

AB - We have applied a new imputation-based method for analyzing incomplete data, called Monte Carlo Bayesian Database Generation (MCBDG), to the spent fuel isotopic composition (SFCOMPO) database. About 60 % of the entries in SFCOMPO are absent. The method estimates missing values of a property from a probability distribution created from the existing data for the property, and then generates multiple instances of the completed database for training a machine learning algorithm. Uncertainty in the data is represented by an empirical or an assumed error distribution. The method makes few assumptions about the underlying data, and it compares favorably against results obtained by replacing missing information with constant values.

KW - Bayesian methods

KW - Machine learning

KW - Missing data

KW - Monte Carlo methods

KW - Nuclear forensics

KW - Spent fuel isotopic composition (SFCOMPO) database

UR - http://www.scopus.com/inward/record.url?scp=84944707514&partnerID=8YFLogxK

U2 - 10.1007/s10967-015-4458-x

DO - 10.1007/s10967-015-4458-x

M3 - Article

AN - SCOPUS:84944707514

SN - 0236-5731

VL - 308

SP - 687

EP - 692

JO - Journal of Radioanalytical and Nuclear Chemistry

JF - Journal of Radioanalytical and Nuclear Chemistry

IS - 2

ER -

Nuclear forensics analysis with missing data

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this