Original scientific paper
METHODOLOGICAL CONSIDERATION ON PRE-PROCESSING DATA OPTIMIZATION CONCERNING AIR DISPERSION MODEL AND NEURAL NETWORKS: A CASE-STUDY OF OZONE PREDICTION LEVEL
A. Pelliccioni
; ISPESL-DIPIA, Via Fontana Candida 1, 00040, Monteporzio Catone (RM), Italy
R. Cotroneo
; ISPESL-DIPIA, Via Fontana Candida 1, 00040, Monteporzio Catone (RM), Italy
F. Pungi
; ISPESL-DIPIA, Via Fontana Candida 1, 00040, Monteporzio Catone (RM), Italy
Abstract
This work analyzes the results of a Neural Network model applied to air pollution data. In particular, we forecast ozone
pollutants levels in a short term using both air dispersion models and neural network methods. The purpose of this work is to provide
a novel methodological procedure to analyze environmental data by using a neural net as forecast technique for ozone levels in the
urban area of Rome. Results show that the model performance can be improved by pre-processing input data using typical datamining
techniques and coupling air dispersion model with neural net.
Keywords
Ozone; neural networks; Data mining; Stepwise algorithm selection; resampling
Hrčak ID:
64321
URI
Publication date:
12.12.2008.
Visits: 913 *