Recognition of Food-Borne Pathogenic Bacteria by Raman Spectroscopy Based on Random Forest Algorithm

Qi Wang; Wandan Zeng; Zhiping Xia; Zhiping Li; Han Qu

doi:10.3788/CJL202148.0311002

Journals >Chinese Journal of Lasers >Volume 48 >Issue 3 >Page 0311002 > Article

Chinese Journal of Lasers
Vol. 48, Issue 3, 0311002 (2021)

Recognition of Food-Borne Pathogenic Bacteria by Raman Spectroscopy Based on Random Forest Algorithm

Qi Wang¹, Wandan Zeng^1、*, Zhiping Xia^2、*, Zhiping Li², and Han Qu²

Author Affiliations

¹College of Computer Science and Information Engineering, Shanghai Institute of Technology, Shanghai 201418, China

²Military Veterinary Institute, Changchun, Jilin 130062, China

show less

DOI: 10.3788/CJL202148.0311002 Cite this Article Set citation alerts

Qi Wang, Wandan Zeng, Zhiping Xia, Zhiping Li, Han Qu. Recognition of Food-Borne Pathogenic Bacteria by Raman Spectroscopy Based on Random Forest Algorithm[J]. Chinese Journal of Lasers, 2021, 48(3): 0311002 Copy Citation Text

show less

Fig. 1. Original Raman spectra

Download full size

Fig. 2. Raman spectra after normalization

Download full size

Fig. 3. Raman spectrum after Savitzky-Golay processing

Download full size

Fig. 4. Pareto chart of principle components

Download full size

Fig. 5. Frame of random forest algorithm

Download full size

Fig. 6. Model accuracy change with n_estimators

Download full size

Fig. 7. Model accuracy change with max_depth

Download full size

Fig. 8. 10-fold cross-validation diagram

Download full size

Number	Latin name
10869	Yersinia enterocolitica
10870	Klebsiella pneumoniae
21482	Salmonella enterica subsp. enterica serovarInfantis
21530	Escherichia coli EHEC O157:H7
21534	Shigella flexneri
21560	Cronobacter sakazakii
21600	Staphylococcus aureus
21617	Vibrio parahaemolyticus
22933	Acinetobacter baumannii
22956	Salmonella enterica subsp. enterica serovarTyphimurium
23794	Vibrio cholerae

Table 1. CICC numbers and names of eleven food-borne pathogenic bacteria

Decision tree algorithm
Input: sample X, sample numbers N, feature counts M
Output:Decision Tree model
X→for bagging∥processing X with bagging cycles
end for
while extracting n_try(n_try=N)→X_train do
M→m_try(m_try≪M)∥ random selection of m_try attributes
m_try→the best node
X→X_samples// build samples using Bootstrap
end while
for (i_tree=0; 1<i_tree≤N_tree; i_tree++)
∥ node splitting by optimal attributes to generate decision trees
end for
end procedure

Table 2. Work process of decision tree algorithm

Food-borne pathogenic bacteria prediction algorithm
Input: sample X, training set X_train, test set X_test
Output: K trees, prediction result r
for all i = 1 to K do
while j≤N do
row_sample=row_sample+select $(X_{train})$
j++
end while
while stop condition not true do
col_sample-select $(ro w_{sample})$
split_Attribute-min{Gini(col_sample)}
∥ classification attributes are determined by the minimum Gini value
tree-AddNode $(split_Attribute)$
end while
leaf_node←node
end for
for all i-1 to K do
R_i=T_i_Predict $(D_{test})$
R=MostCommon(R_i)
end for
end procedure