Search

This paper proposed data mining techniques to improve efficiency and reliability in diabetes classification. The real data set collected from Sawanpracharak Regional Hospital, Thailand, was fist analyzed by using gain-ratio feature selection techniques. Three well known algorithms; naïve bayes, k-nearest neighbors and decision tree, were used to construct classification models on the selected features. Then, the popular ensemble learning; bagging and boosting were applied using the three base classifiers. The results revealed that the best model with the highest accuracy was bagging with base classifier decision tree algorithm (95.312%). The experiments also showed that ensemble classifier models performed better than the base classifiers alone.

Polybrominated biphenyls (PBBs) and polybrominated biphenyl ethers (PBDEs) are substances used in electrical and electronic products. They are restricted substances due to their potential risk to human health and the environment. In this research work, a test method was evaluated for the quantitative analysis of polybrominated diphenyl ethers (PBDEs) and polybrominated biphenyls (PBBs) in acrylonitrile-butadiene–styrene (ABS) plastic and polyethylene materials, based on solvent extraction of a plastic sample with soxhlet extraction techniques and detection by gas chromatography/mass spectrometry. We have also developed a computer program for checking the suitability of the analytical method by evaluating validation characteristics, including accuracy, precision, specificity/selectivity, limit of detection (LOD), limit of quantification (LOQ), linearity, range and sensitivity. The computer program was successfully applied for evaluation of the analytical method for prohibited PBBs and PBDEs in plastics. Graphical charts for the evaluation of the analytical method are presented along with their values, and researchers can get insight into evaluating several parameters together. It was found that the analytical method was able to quantify monobrominated - decabrominated biphenyl and monobrominated diphenyl ether–nonabrominated diphenyl ether between 100-1750 mg/kg and decabrominated diphenyl ether between 100–8,000 mg/kg. In addition, ERM-EC590 Polyethylene certified reference material from the Institute for Reference Materials and Measurements (IRMM) was used to validate the analytical method for PBBs and PBDEs in plastics. The obtained accuracy and precision indicate that the method successfully meets the required accuracy expressed as % recovery (70-130%) and the required precision expressed as percentage relative standard deviation (less than 20% RSD) of the IEC 62321:2008 international standard method.

This paper discusses how Random Forests, ensembles of weak decision trees, can be improved by excluding less important features from the model. Gain Ratio Feature Selection was used as the basis for tuning the algorithm parameters. Backwards elimination of the features to obtain the minimum subset with the highest accuracy was the key methodology of this experiment. The results of the proposed model were better in terms of accuracy and number of features used. The objective of this paper was to create a base-line, which will be useful for the classification on diabetes complications data. We recommend using the Random Forest with Feature Selection technique for other type of classification problems. Future work also includes an extension study of the different types of learning settings to improve the feature construction process.