Sylvio Barbon Jr., Ana Paula A. C. Barbon, Rafael Gomes Mantovani, and Douglas Fernande Barbin
Decision Table, Decision Tree, Quality, M5P
Classification of poultry meat quality relies on inconsistent, time-consuming, and laborious techniques often conducted subjectively. Measurement of attributes such as water holding capacity (WHC) and interpretation of other parameters such as pH and colour features depends on trained personnel and specific equipments for each test. The current work presents a combined attribute selection step and classification step for poultry meat samples. Samples from different quality attributes were analysed to comprise a large set of values for each parameter. REPTRee decision trees predictor
exploit the optimal parameters for classification tasks
of different quality grades of poultry meat. The proposed
methodology was conducted with a Support Vector Machine algorithm (SVM) with standard parameters to compare model precision without a pipeline of processes. Experiments were performed on colour (CIEL*a*b*, chroma and hue), water holding capacity (WHC), and pH of each sample analysed. Results show that the best method was a REPTree based on 2 parameters, allowing for classification of poultry samples on quality grades with 98:1% precision.