Identification of Boletus Species Based on Discriminant Analysis of Partial Least Squares and Random Forest Algorithm

Feng-xia CHEN; Tian-wei YANG; Jie-qing LI; Hong-gao LIU; Mao-pan FAN; Yuan-zhong WANG

doi:10.3964/j.issn.1000-0593(2022)02-0549-06

Journals >Spectroscopy and Spectral Analysis >Volume 42 >Issue 2 >Page 549 > Article

Spectroscopy and Spectral Analysis
Vol. 42, Issue 2, 549 (2022)

Identification of Boletus Species Based on Discriminant Analysis of Partial Least Squares and Random Forest Algorithm

Feng-xia CHEN^1、*, Tian-wei YANG^2、2;, Jie-qing LI^1、1;, Hong-gao LIU^3、3;, Mao-pan FAN^{1、1; *;}, and Yuan-zhong WANG^{4、4; *;}

Author Affiliations

¹1. College of Resources and Environmental Sciences, Yunnan Agricultural University, Kunming 650201, China

²2. Yunnan Institute for Tropical Crops Research, Jinghong 666100, China

³3. College of Agronomy and Biotechnology, Yunnan Agricultural University, Kunming 650201, China

⁴4. Institute of Medicinal Plants, Yunnan Academy of Agricultural Sciences, Kunming 650200, China

show less

DOI: 10.3964/j.issn.1000-0593(2022)02-0549-06 Cite this Article

Feng-xia CHEN, Tian-wei YANG, Jie-qing LI, Hong-gao LIU, Mao-pan FAN, Yuan-zhong WANG. Identification of Boletus Species Based on Discriminant Analysis of Partial Least Squares and Random Forest Algorithm[J]. Spectroscopy and Spectral Analysis, 2022, 42(2): 549 Copy Citation Text

show less

Boletus samples(a): Boletus griseus Frost; (b): Boletus edulis Bull.: Fr;(c): Boletus umbriniporus Hongo; (d): Boletus speciosus Forst;(e): Leccinum rugosicepes (Perk) Sing;(f): Boletaceae bicolor Peck; (g): Boletus tomentipes Earle

Fig. 1. Boletus samples
(a): Boletus griseus Frost; (b): Boletus edulis Bull.: Fr;(c): Boletus umbriniporus Hongo; (d): Boletus speciosus Forst;(e): Leccinum rugosicepes (Perk) Sing;(f): Boletaceae bicolor Peck; (g): Boletus tomentipes Earle

Download full size | View in the Article

Fig. 2. Average spectra of 7 species of Boletus
(a): Mid-infrared spectrum; (b): UV-Vis spectrum

Download full size | View in the Article

Fig. 3. Ntree selection diagram and Correct rate matrix
(a): Mid-level (LVs) Ntree best choice map; (b): Mid-level (CPA) Ntree best choice map;(c): Mid-level (LVs) Training set correct rate matrix; (d): Mid-level (CPA) Training set correct rate matrix;(e): Mid-level (LVs) Validation set correct rate matrix; (f): Mid-level (CPA) Validation set correct rate matrix

Download full size | View in the Article

种类	编号	产地	数量
灰褐牛肝菌	a	保山市隆阳区、玉溪市江川区、大理市弥渡县、昆明市晋宁区、曲靖市马龙区、昆明市安宁区	98
美味牛肝菌	b	楚雄市南华县、迪庆州香格里拉市、大理市弥渡县、迪庆州维西县、保山市隆阳区、昆明市安宁区、玉溪市红塔区、大理市鹤庆县、文山市东山乡、昆明市石林县	221
栗色牛肝菌	c	昆明市石林县、曲靖市马龙区、保山市隆阳区、大理市弥渡县、红河州石屏县、楚雄市元谋县、玉溪市红塔区、红河州个旧市	110
小美牛肝菌	d	楚雄市南华县、玉溪市红塔区、保山市隆阳区	32
皱盖疣柄牛肝菌	e	曲靖市麒麟区、红河州石屏县、大理市弥渡县、昆明市安宁区、迪庆州维西县、保山市隆阳区、楚雄市元谋县、玉溪市红塔区	132
双色牛肝菌	f	楚雄市南华县、大理市苍山、曲靖市麒麟区	30
绒柄牛肝菌	g	红河州个旧市、迪庆州香格里拉市、红河州石屏县、大理市鹤庆县	60

Table 1. Boletus samples information

View in the Article

光谱类型	预处理方法	主成分数	${R^{2 Y}}_{cum}$	$Q_{cum}^{2}$	RMSEC	RMSECV	训练集/%	验证集/%
中红外光谱	SNV+SG	24	0.693	0.555	0.172 449	0.211 937	97.59	98.24
	2D+MSC+SNV	18	0.778	0.651	0.149 082	0.193 025	99.78	99.12
	1D+MSC+SNV+SG	17	0.673	0.575	0.178 289	0.210 069	97.37	97.80
紫外光谱	MSC+2D	7	0.038 2	0.024 7	0.320 725	0.329 83	34.65	32.16
	SNV+SG	16	0.426	0.281	0.247 129	0.282 096	80.70	80.18
	2D+MSC+SNV	10	0.351	0.262	0.265 682	0.284 48	67.32	65.64
	1D+MSC+SNV+SG	16	0.364	0.257	0.262 953	0.284 499	76.10	73.13
低级融合	2D+MSC+SNV, SNV+SG	19	0.8	0.676	0.140 484	0.185 499	100	99.12
中级融合(LVs)		6	0.834	0.812	0.131 507	0.148 948	100	99.56
中级融合(CPA)		11	0.927	0.882	0.086 528	0.152 926	100	100

Table 2. The main parameters and accuracy of the discriminant analysis model of partial least squares

View in the Article

光谱类型	预处理方法	Ntree	Mtry	训练集/%	验证集/%
中红外光谱	SNV+SG	2 000, 1 412	40, 3	76.75	85
	2D+MSC+SNV	1 800, 1 149	50, 5	93.20	99
	1D+MSC+SNV+SG	2 100, 1 537	42, 5	87.72	96.29
紫外光谱	2D+MSC	2 000, 1 821	40, 4	62.06	64.30
	SNV+SG	1 700, 1 327	35, 2	56.80	70.71
	2D+MSC+SNV	2 000, 1 549	40, 5	64.47	73.10
	1D+MSC+SNV+SG	1 800, 1 562	30, 4	62.28	76.14
低级融合	2D+MSC+SNV, 1D+MSC+SNV+SG	2 000, 1 184	40, 5	92.32	99.14
中级融合(LVs)		2 000, 865	40, 5	92.76	96.04
中级融合(CPA)		2 000, 1 249	40, 5	97.15	100

Table 3. The main parameters and accuracy of the random forest model

Download Citation

Save the article for my favorites

Paper Information