Technical Paper

Data Estimation and Distribution Map Using CQSAR: Compensation Quantitative Structure-Activity Relationships

Junko KAMBEa, Umpei NAGASHIMAb,c* and Tomoo AOYAMAd

aFaculty of Media Communication, Edogawa University
474 Komaki, Nagareyama, Chiba 270-0198, Japan
bResearch Institute for Computational Science (RICS), National Institute of Advanced Industrial Science and Technology (AIST)
1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan
cCREST, Japan Science and Technology Agency
4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan
dFaculty of Technology, Miyazaki University
Gakuenkihanadai Nishi, Miyazaki 889-2192, Japan

(Received: May 1, 2008; Accepted for publication: September 16, 2008; Advance publication: October 24, 2008)

In this note, we present a method to draw a distribution map of materials observed at a few, sparse and localized data points, a situation frequently appeared in the problem of chemistry and environmental science. In this situation, it is usually difficult to draw a distribution map because data on triangular or square lattice are required. We generate additional points where the amount of materials is estimated by CQSAR: Compensation Quantitative Structure-Activity Relationships based on perceptron type neural network. Then the data were used for drawing a distribution map. The distribution map using the additional points generated by CQSAR is useful for understanding an overview of material distributions.

Keywords: NOx, Suspended Particulate Matter, Distribution Map, CQSAR, Neural Network

Abstract in Japanese

Text in Japanese(HTML)

PDF file on J-STAGE