Bioscience of Microbiota, Food and Health
Online ISSN : 2186-3342
ISSN-L : 2186-3342
Full Paper
Technical Aspects of Nominal Partitions on Accuracy of Data Mining Classification of Intestinal Microbiota — Comparison between 7 Restriction Enzymes
Toshio KOBAYASHIKenji FUJIWARA
Author information
JOURNALS FREE ACCESS

2014 Volume 33 Issue 3 Pages 129-138

Details
Abstract

The application of data mining analyses (DM) is effective for the quantitative classification of human intestinal microbiota (HIM). However, there remain various technical problems that must be overcome. This paper deals with the number of nominal partitions (NP) of the target dataset, which is a major technical problem. We used here terminal restriction fragment length polymorphism data, which was obtained from the feces of 92 Japanese men. Data comprised operational taxonomic units (OTUs) and subject smoking and drinking habits, which were effectively classified by two NP (2-NP; Yes or No). Using the same OTU data, 3-NP and 5-NP were examined here and results were obtained, focusing on the accuracies of prediction, and the reliability of the selected OTUs by DM were compared to the former 2-NP. Restriction enzymes for PCR were further affected by the accuracy and were compared with 7 enzymes. There were subjects who possess HIM at the border zones of partitions, and the greater the number of partitions, the lower the obtained DM accuracy. The application of balance nodes boosted and duplicated the data, and was able to improve accuracy. More accurate and reliable DM operations are applicable to the classification of unknown subjects for identifying various characteristics, including disease.

Information related to the author
© 2014 by BMFH Press

This article is licensed under a Creative Commons [Attribution-NonCommercial-NoDerivatives 4.0 International] license.
https://creativecommons.org/licenses/by-nc-nd/4.0/
Previous article
feedback
Top