3.2.1. SIMCA
Table 3 shows the optimal PC number employed for modeling of each class and the two types of classification errors: (a) a type-I error, which consists of a sample not included in its own class; and (b) a type-II error, which consists of a sample included in an incorrect class. We can observe that all samples were correctly classified into their respective class for all used data sets, except for the Argentinean green teas when using the ‘‘AFC parameters’’ that had a classification error of 20%. In relation to the type II errors, the best SIMCA result was achieved by using the ‘‘Elements in tea infusions’’ data set, which reached a mean correct classification of 88.9%. The other ‘‘All parameters’’, ‘‘AFC parameters’’, and ‘‘Total content of elements’’ data sets presented mean classification errors of 25, 61.1 and 71.7%, respectively.
3.2.1. SIMCATable 3 shows the optimal PC number employed for modeling of each class and the two types of classification errors: (a) a type-I error, which consists of a sample not included in its own class; and (b) a type-II error, which consists of a sample included in an incorrect class. We can observe that all samples were correctly classified into their respective class for all used data sets, except for the Argentinean green teas when using the ‘‘AFC parameters’’ that had a classification error of 20%. In relation to the type II errors, the best SIMCA result was achieved by using the ‘‘Elements in tea infusions’’ data set, which reached a mean correct classification of 88.9%. The other ‘‘All parameters’’, ‘‘AFC parameters’’, and ‘‘Total content of elements’’ data sets presented mean classification errors of 25, 61.1 and 71.7%, respectively.
การแปล กรุณารอสักครู่..
