DISCOVERY OF RELATION BETWEEN Listeria AND OTHER BACTERIAL CONTAMINATION USING CLASSIFICATION REFINEMENT TECHNIQUE

Abstract

People often use chicken to prepare their diary meal which creates a great demand for food producers and exporters in Thai food industry. Thailand is ranked fourth in chicken exporters in the world. Standards on chicken meat regulations differ from country to country. One commonality remains: controlling the bacteria that can affect human life, particularly pregnant women and the unborn. One of the important bacteria is called Listeria which receives high attention for the industry to prevent their contamination in chicken. A microbiological test is usually conducted to analyze the data from industry. However, the microbiological test is incapable of identifying other variants from Listeria contamination. There is no computer model that can be used in Listeria contamination analysis. Anyhow, such tests create plenty of data that can be applied with a computer model to investigate the factors that causes Listeria contamination. This research proposes a classification refinement technique which composes of random forest and linear regression to predict the variants of Listeria contamination. Moreover, Naïve Bayes is used to assess the model correctness. This proposed procedure can reveal the important causes relating to variants and the conditions on the number of these variants that produce Listeria contamination.