Ellea General percent C. cibarius B. Choline (bitartrate) Cancer edulis A. mellea General percent Predicted A. mellea 0 1 five ten.35 0 1 two 15.79C. cibarius 27 1 2 51.73 six 0 0 31.58B. edulis 1 18 three 37.94 0 ten 0 52.64Percent Correct 96.43 90.00 50.00 86.21 one hundred 90.91 one hundred 94.74TrainingHoldoutIn the instruction step, the overall % of correctly classified samples was 86.21 , although for the holdout set, the % rose to 94.74 . The reduce values obtained for the instruction set had been on account of one C. cibarius sample, two B. edulis, and 5 A. mellea, which have been placed to other species. In any case, for the holdout sample, only a single sample from B. edulis was misclassified. With regards to the options selection, only 3 points were selected: 1746 cm-1 , 1510 cm-1 , and 1388 cm-1 . The samples’ distribution in between the two sets, in accordance with chosen options, is presented in Figure 3 below: It needs to be noticed that the outcomes obtained employing PCA-LDA and kNN are very comparable with regards to species prediction accuracy. Regarding the obtained predictors, it must be talked about that, except for 1746 cm-1 , which also appeared in LDA classification, the other two bands are new predictors. This could lead to the conclusion that these two approaches are complementary. The amount of groups for fuzzy c-means clustering (FCM) analysis was chosen according to the three investigated species, namely 3. The sample codes for this evaluation have been as follows: code 1 for Armillaria mellea (samples 12), code 2 for Boletus edulis (samples 133), and code three for Cantharellus cibarius (samples 447). FCM made threeTrainingAppl. Sci. 2021, 11,HoldoutB. edulis A. mellea All round percent C. cibarius B. edulis A. mellea Overall percent1 two 51.73 6 0 0 31.5818 three 37.94 0 ten 0 52.641 5 ten.35 0 1 two 15.7990.00 50.00 86.21 100 90.91 one hundred 94.747 offuzzy partitions, which have been all represented by a prototype (a cluster center with the specIn the coaching step, the all round percent of appropriately classified samples was 86.21 , trum corresponding towards the fuzzy robust means on the original FT-IR spectra traits while for the holdout set, the % rose to 94.74 . The lower values obtained for the for 77 samples weighted by degree of membership (DOM)) corresponding to every single partition. education set had been as a consequence of one C cibarius sample, two B. edulis, and 5 A. mellea, which To compare the partitions, the similarities and differences amongst samples, the spectra of the had been placed to other species. In any case, for the holdout sample, only 1 sample from prototypes corresponding towards the 3 fuzzy partitions (A1 three) obtained by applying both B. edulis and DOMs of samples corresponding to all fuzzy partitions, need to be analyzed. The FCM was misclassified. Concerning the characteristics choice, only 3 points had been se-1 lected: 1746 cm-1, 1510Table, 2and 1388 cm-1clearly illustratedistribution among the two results presented in cm and Figure four . The samples’ one of the most distinct traits sets, according to selected and their is presented in Figure 3sample assignment based on attributes, (dis)similarity as well as the under: of every single fuzzy partition their DOMs.Figure 3. kNN modeling of SS-208 Purity mushroom samples, with 3 options chosen and five neighbors. Figure 3. kNN modeling of mushroom samples, with three attributes chosen and 5 neighbors. Table two. The 3 fuzzy partitions obtained by applying the fuzzy c-means clustering technique.Fuzzy Partition A A1 A2 A1, 10, 13, 15, 16, 17, 18, 19, 20, 21, 22, 23, two.
