Statistical prediction models for the odour emissions quantification in terms of odour concentration: Analysis and comparison

Measuring odour concentration is a significant step to achieve efficient environmental odour management in continuous, objective and repeatable manner. To deal with this, researchers developed instrumental odour monitoring systems (IOMS) by applying odour monitoring models (OMM) for prediction. At present, limited data are available in the literature regarding the exploration of different prediction models to quantify the odour emissions in terms of odour concentration. This study presents and compares different types of parametric and nonparametric predictive models (i.e., artificial neural network (ANN), multivariate adaptive regression splines (MARSpline), partial least squares (PLS), multiple linear regression (MLR), response surface regression (RSR)) with the aim to increase the reliability of the odour concentration prediction by using IOMS for environmental odour monitoring. The experimental studies are carried out considering odour samples collected from the organic fractions in municipal solid waste. All samples undergone seedOA eNose and dynamic olfactometry analysis as reference methods. The coefficient of determination (R2) and root mean square error (RMSE) were used to measure the goodness-of-fit of the models. Results indicate the strengths and weaknesses of the analyzed models and highlight their accuracy in terms of odour concentration prediction.