JURNAL RISET INFORMATIKA
Vol. 5, No. 2 March 2022
P-ISSN: 2656-1743 |E-ISSN: 2656-1735
DOI: https://doi.org/10.34288/jri.v5i2.467
Accredited rank 3 (SINTA 3), excerpts from the decision of the Minister of RISTEK-BRIN No. 200/M/KPT/2020
171
EVALUATION OF MACHINE LEARNING USING THE K-NN ALGORITHM
TO DETERMINE THE QUALITY OF MEAT BEFORE CONSUMPTION
Feronika-1*), Masrizal-2, Ibnu Rasyid Munthe-3
1,3 Department of Information system, 2Department of Informatics Management
1,2,3 Universitas Labuhanbatu
Sumatera Utara, Indonesia
1*) feronika@gmail.com, 2masrizal120405@gmail.com, 3ibnurasyidmunthe@gmail.com
(*) Corresponding Author
Abstract
Meat is one of the sources of animal protein for humans, and one of the requirements that must be met so
that the human body does not lack protein, especially animal; this protein can be obtained from beef,
chicken, and other meats, but the most important thing here is the content contained in meat, whether it
has been contaminated with chemicals, e.g., chicken that has been injected with chemicals that cause the
chicken to look fat, or beef whose flexibility has decreased and the pH is getting more acidic. This research
tries to predict meat quality by looking at two parameters: flexibility and acidity. The programming
language used is R Language, using the k-NN method or Algorithm to determine the meat's condition
suitable for consumption. In detail, it will be processed in Machine Learning using the k-NN Algorithm; there
are two criteria for consumption of meat, namely good or not good for consumption; in detail, the output
will be explained using a specific graph using a plot function, and array data will be specifically classified to
represent values. The value of 2 variables, namely feasible or not suitable for consumption.
Keywords: Machine Learning, K-NN Algorithm, R Language, Meat, Acidity Prediction, Flexibility Prediction
Abstrak
Daging adalah salah satu sumber protein hewani bagi manusia, dan salah satu syarat yang harus dipenuhi
agar tubuh manusia tidak kekurangan protein khususnya hewani, protein ini bisa didapatkan dari sapi,
ayam, dan daging lainnya, namun yang terpenting disini adalah kandungan yang terdapat pada daging,
apakah sudah terkontaminasi dengan zat kimia, misalnya ayam potong yang telah disuntik dengan zat kimia
yang menyebabkan ayam kelihatan gemuk, atau daging sapi yang tingkat kelenturannya sudah berkurang
dan pH yang semakin asam. Riset ini mencoba untuk melakukan prediksi kualitas daging dengan melihat
dua parameter yaitu kelenturan dan keasaman. Bahasa pemrograman yang digunakan adalah R Languag e,
menggunakan metode atau algoritma k-NN yang dapat menentukan kondisi daging layak untuk dikonsumsi.
Secara detail akan diolah pada Machine Learning menggunakan algoritma k-NN ini, terdapat dua kriteria
daging konsumsi yaitu bagus atau kurang bagus untuk dikonsumsi, secara detail output akan dijelaskan
menggunakan specific graph menggunakan plot function, dan data array akan secara spesifik
diklasifikasikan untuk merepresentasikan nilai-nilai dari 2 variable yaitu layak atau tidak layak konsumsi.
Kata kunci: Machine Learning, K-NN Algorithm, R Language, Daging, Prediksi Keasaman, Prediksi
Kelenturan
INTRODUCTION
The economic factor is an essential and
decisive factor to be able to escape the recession in
Indonesia in 2023, so what needs to be done by
farmers is to utilize the existing land for several
types of livestock, including broiler breeders. Beef
or chicken in the freezer for a long time causes the
level of flexibility in the beef to be reduced or hard.
Likewise, with cattle farms, broiler breeders, and
other breeders such as rabbits, goats, and others
whose type of cuisine tends to be "Sate."
Researchers need to write and share in detail new
methods such as AIoT & LoRaWAN (Adi & Kitagawa,
2020), (Liani et al., 2021), (Mukti et al., 2021).
Machine Learning and k-NN Algorithm (Lv et al.,
2021), (Du & Li, 2019), ((Jia, 2022) in solving
quality problems of beef, chicken, and other types
of meat consumption. One of the Indonesian
people's favorite foods is 'Meatballs', like a resident
of Malang, East Java, Indonesia, is one of the fans of
meatballs with many variants of meatballs.
P-ISSN: 2656-1743 | E-ISSN: 2656-1735
DOI: https://doi.org/10.34288/jri.v5i2.467
JURNAL RISET INFORMATIKA
Vol. 5, No. 1 March 2022
Accredited rank 3 (SINTA 3), excerpts from the decision of the Minister of RISTEK-BRIN No. 200/M/KPT/2020
172
Meatballs are also made from meat, e.g., beef.
However, if when it is processed into meatballs, you
see the quality of the beef being grilled, is it healthy,
fresh, or the condition of the meat is no longer red
or bluish.
Figure 1. k-NN Algorithm overview
Figure 1 can be seen as an example of a binary
classification with k=3. The green dot in the middle
is the test sample x. The labels of the three
neighbors are 2×(+1), and 1×(-1) yields the
majority prediction (+1). K-NN Algorithm
((Pawlovsky & Matsuhashi, 2017),(Palacios &
Suzuki, 2019) uses a distance matrix, and it
concludes that the closest parameter is the best
conclusion. Equation 1 shows the formula for this
K-NN Algorithm (Chang & Liu, 2011), (Yao & Cao,
2022), (Salim et al., 2020), (Song et al., 2020), (Li et
al., 2020).
dist(x,x’)≥ max dist(x,x’’), (x’’,y’’) ⋲ 𝑆𝑥 ................... (1)
dist(x, z)=(∑ |𝑋𝑟 − 𝑍𝑟 |
𝑑
𝑟=1
𝑝
)
1/𝑝
................................. (2)
h(x)=mode ({y”( x’’,y’’) ⋲ 𝑆𝑥 }) .................................. (3)
√∑ 𝑥𝑖 − 𝑦𝑖
2𝑘
𝑖=1 .................................................................. (4)
Figure 2. k-NN Algorithm for the feasibility of meat consumption
Furthermore, Figure 2 explains the method used in
this manuscript where the k-NN method (Yao &
Cao, 2022), (Song et al., 2020), (Sun et al., 2020),
(Sushmitha & Jagadeesh, 2022) can solve the
problem of Eligible or Not Like Meat consumed by
the community, especially in Indonesia.
RESEARCH METHODS
This research will focus on using the k-NN
Algorithm (Zhai, 2022), (Li et al., 2020), (Jia, 2022),
(Sushmitha & Jagadeesh, 2022), (Sun et al., 2020),
(Yunneng, 2020), (Palacios & Suzuki, 2019) and
(Setia & Garg, 2021). This Algorithm is used to
classify new objects based on attributes and
samples from training data, and this Algorithm uses
the predicted value of the new instance value.
Step-by-step to running of k-NN Algorithm in general:
1. Choose k nearest neighbors randomly
2. Map dataset to vector space
3. Separate the dataset into training data and test
data
4. Calculate the distance, d, between the test data
and the training data
5. Sorting d from smallest to largest
6. Take and separate k-sorting data
7. Observe the majority class
8. Classify test data by majority
JURNAL RISET INFORMATIKA
Vol. 5, No. 2 March 2022
P-ISSN: 2656-1743 |E-ISSN: 2656-1735
DOI: https://doi.org/10.34288/jri.v5i2.467
Accredited rank 3 (SINTA 3), excerpts from the decision of the Minister of RISTEK-BRIN No. 200/M/KPT/2020
173
Furthermore, it will be explained in full in the
following flowchart, where the value of k must be
determined first, then calculate the distance, sort it,
and classified the test data. Moreover, Pseudo code
is made to understand the flow of the program so
that readers do not experience difficulties in
understanding (Setia & Garg, 2021).
Figure 3. k-NN Algorithm flowchart
Pseudocode_1
// X: training data
// Y: class labels of X
//x: unknown sample
Classify (X,Y,x)
For i = 1 to m do
Compute distance d(X,..,x)
end for
Compute set 1 containing indicates for the k smallest
distances d (X,...x).
End for
Compute set 1 containing indicates for the k smallest
distances d (X,..x).
Return majority label for {Yi where I e I}
RESULTS AND DISCUSSION
The following table data determines the value to
look for or predictions that will give the best quality
value on meat quality. Moreover, Table 1 is a
Classification of Meat Quality.
Table 1. Classification of Meat Quality
Brand acidity
value
Flexibility value Category
A 7 1.2 Good
B 6 1.7 Not good
C 8 1.5 Good
D 5 1.3 Not good
E 9 1.0 Good
F 9.5 1.4 Good
G 8.3 0.8 Not good
H 7.5 1.1 ?
Furthermore, the next step is to run the program
using R Language using K-NN Algorithm (Salim et
al., 2020), (Riquelme et al., 2020); the display on the
Website page is data taken; this is stage 1. next is
The relationship between flexibility and uniformity
of product data as specifically shown in figure 4.
Code_1_running
x<-c(7,6,8,5,9,9.5,8.3,7.5)
> y<-
c(1.2,1.7,1.5,1.3,1.0,1.4,0.8,1.1)
> bagus_x<-c(7,8,9,9.5)
> bagus_y<-c(1.2,1.5,1.0,1.4)
> kurang_x<-c(6,5,8.3)
> kurang_y<-c(1.7,1.3,0.8)
> test_x<-(7.5)
> test_y<-(1.1)
> merk_bagus<-c('A','C','E','F')
> merk_kurang<-c('B','D','G')
> merk_test<-c('H')
>plot(x,y,col="blue",main="dataproduk"
,cex=1.3,pch=16,xlab="keasamaan",ylab=
"kelenturan")
Step-by-step to running of k-NN Algorithm on this
project:
1. Group product data into product array objects
according to their respective categories
2. Group the product brand data (label) into the
brand array object according to their respective
categories
3. Plot an Array of products using different symbols
4. Plot array brands
5. Plot legend
P-ISSN: 2656-1743 | E-ISSN: 2656-1735
DOI: https://doi.org/10.34288/jri.v5i2.467
JURNAL RISET INFORMATIKA
Vol. 5, No. 1 March 2022
Accredited rank 3 (SINTA 3), excerpts from the decision of the Minister of RISTEK-BRIN No. 200/M/KPT/2020
174
Figure 4. The relationship between flexibility and
uniformity of product data
Step 2 is to determine the value of k; for example, if
k is 3, then k=3, we can choose k odd, 1, 3, or 5. Then
step 3 is to calculate the distance (d) between the
test data (H) and neighbors; based on the results of
observations of the type of data and the shape of the
graph plot, then we can use the Euclidean distance
formula to determine the distance d as the Figure 5.
Euclidean distance (d) = √(𝑥2 − 𝑥1)
2 + (𝑦2 − 𝑦1)
2
Figure 5. The Euclidean distance (d)
Then we will determine the coordinate value, the
coordinate value is seen in Table 2.
Table 2. Coordinate value
Number Coordinate
1 A(7, 1.2)
2 B (6, 1.7)
3 C(8,1.5)
4 D(5,1.3)
5 E(9,1.0)
6 F (9.5,1.4)
7 G(8.3,0,8)
8 H(7.5,1.1)
𝑑𝐻𝐴 = √7.5 − 7
2 + 1.1 − 1.22 = 0.509902
𝑑𝐻𝐵 = √7.5 − 6
2 + 1.1 − 1.72 = 1.615549
𝑑𝐻𝐶 = √7.5 − 8
2 + 1.1 − 1.52 = 0.640312
𝑑𝐻𝐷 = √7.5 − 5
2 + 1.1 − 1.32 = 2.507987
𝑑𝐻𝐸 = √7.5 − 9
2 + 1.1 − 1.02 = 1.503333
𝑑𝐻𝐹 = √7.5 − 9.5
2 + 1.1 − 1.42 = 2.02237
𝑑𝐻𝐺 = √7.5 − 8.3
2 + 1.1 − 0.82 = 0.85440
Stage 4 is sorting the results of the d calculations
from the smallest to the most significant d. Then
choose d as much as the value of k, namely k = 3
pieces, before sorting the results of the d calculation
as following Table 3.
Table 3. Distance calculation
d d value Classification
HA 0.509902 A (Good)
HB 1.615549 B (Good)
HC 0.640312 C (Not good)
HD 2.507987 D (Not good)
HE 1.503333 E (Not good)
HF 2.022375 F (Good)
HG 0.854400 G (Good)
The next step is to determine a good product with a
cross and determine the test data. The plot results
can be seen in Figure 3. In Figure 3, those marked
with a cross are of good quality, while those not
marked with a cross are data that are not of good
quality. The box is a sample of test data. Moreover,
The relationship between flexibility and uniformity
of product data is presented explicitly in Figure 6.
Code_2_running
>points(bagus_x,bagus_y,col="red",pch=
4,lwd=2,cex=2)
>points(test_x,test_y,col="black",pch=
22,lwd=2,cex=2)
Figure 6. The relationship between flexibility and
uniformity of product data
CONCLUSIONS AND SUGGESTIONS
Conclusion
Using the K-NN Algorithm using R
Language, the meat quality can be determined by
several parameters, namely the value of flexibility
and the value of acidity, which is inputted for
processing or the data to be trained by the K-NN
Algorithm. Good quality of the meat was found at
acidity seven and flexibility 1.2, 8 and 1.5, 9 and 1.0,
and 9.5 and 1.4
Suggestion
The new method can be compared to
produce an error comparison value (%) of the
quality of processed meat or meat consumed by the
JURNAL RISET INFORMATIKA
Vol. 5, No. 2 March 2022
P-ISSN: 2656-1743 |E-ISSN: 2656-1735
DOI: https://doi.org/10.34288/jri.v5i2.467
Accredited rank 3 (SINTA 3), excerpts from the decision of the Minister of RISTEK-BRIN No. 200/M/KPT/2020
175
community if it is seen not only as two parameters
but can be more.
REFERENCES
Adi, P. D. P., & Kitagawa, A. (2020). Performance
evaluation of low power wide area (LPWA)
LoRa 920 MHz sensor node to medical
monitoring IoT based. EECCIS 2020 - 2020
10th Electrical Power, Electronics,
Communications, Controls, and Informatics
Seminar, 278–283.
https://doi.org/10.1109/EECCIS49483.2020
.9263418
Chang, Y., & Liu, H. (2011). Semi-supervised
classification algorithm based on the KNN.
2011 IEEE 3rd International Conference on
Communication Software and Networks, ICCSN
2011, 9–12.
https://doi.org/10.1109/ICCSN.2011.60143
76
Du, S., & Li, J. (2019). Parallel Processing of
Improved KNN Text Classification Algorithm
Based on Hadoop. 2019 7th International
Conference on Information, Communication
and Networks (ICICN), 167–170.
https://doi.org/10.1109/ICICN.2019.883497
3
Jia, X. (2022). Detecting Water Quality Using KNN,
Bayesian and Decision Tree. Proceedings -
2022 Asia Conference on Algorithms,
Computing and Machine Learning, CACML
2022, 323–327.
https://doi.org/10.1109/CACML55074.2022
.00061
Li, B., Wan, S., Xia, H., & Qian, F. (2020). The
Research for Recommendation System Based
on Improved KNN Algorithm. Proceedings of
2020 IEEE International Conference on
Advances in Electrical Engineering and
Computer Applications, AEECA 2020, 796–798.
https://doi.org/10.1109/AEECA49918.2020.
9213566
Liani, Y. A., Munthe, I. R., Irmayani, D., Broto, B. E.,
Yanris, G. J., Prasetya, D. A., Haryanto, R., Adi,
P. D. P., Muslikh, A. R., & Arifuddin, R. (2021).
The Broiler Chicken Coop Temperature
Monitoring Use Fuzzy Logic and LoRAWAN.
Proceeding - ICERA 2021: 2021 3rd
International Conference on Electronics
Representation and Algorithm, 161–166.
https://doi.org/10.1109/ICERA53111.2021.
9538771
Lv, W., Huang, H., Tang, W., & Chen, T. (2021).
Research and Application of Intersection
Similarity Algorithm Based on KNN
Classification Model. 2021 International
Conference on Artificial Intelligence, Big Data
and Algorithms (CAIBDA), 141–144.
https://doi.org/10.1109/CAIBDA53561.202
1.00037
Mukti, F. S., Adi, P. D. P., Prasetya, D. A., Sihombing,
V., Rahanra, N., Yuliawan, K., & Simatupang, J.
(2021). Integrating Cost-231 Multiwall
Propagation and Adaptive Data Rate Method
for Access Point Placement Recommendation.
International Journal of Advanced Computer
Science and Applications, 12(4), 772–777.
https://doi.org/10.14569/IJACSA.2021.0120
494
Palacios, P. A., & Suzuki, Y. (2019). An Immune
Algorithm that Uses a Master Cell for
Component Selection for the kNN Method.
Pan American Health Care Exchanges, PAHCE,
2019-March.
https://doi.org/10.1109/GMEPE-
PAHCE.2019.8717318
Pawlovsky, A. P., & Matsuhashi, H. (2017). The use
of a novel genetic algorithm in component
selection for a kNN method for breast cancer
prognosis. 2017 Global Medical Engineering
Physics Exchanges/Pan American Health Care
Exchanges (GMEPE/PAHCE), 1–5.
https://doi.org/10.1109/GMEPE-
PAHCE.2017.7972084
Riquelme, J. A., Barrientos, R. J., Hernandez-Garcia,
R., & Navarro, C. A. (2020). An exhaustive
algorithm based on GPU to process a kNN
query. 2020 39th International Conference of
the Chilean Computer Science Society (SCCC),
1–8.
https://doi.org/10.1109/SCCC51225.2020.9
281231
Salim, A. P., Laksitowening, K. A., & Asror, I. (2020).
Time Series Prediction on College Graduation
Using KNN Algorithm. 2020 8th International
Conference on Information and
Communication Technology, ICoICT 2020.
https://doi.org/10.1109/ICOICT49345.2020.
9166238
Setia, V., & Garg, S. (2021). Psychometric
Assessment of Teacher Mindset using hybrid
KNN-SVM Algorithm. Proceedings of the 5th
International Conference on I-SMAC (IoT in
Social, Mobile, Analytics and Cloud), I-SMAC
2021, 951–955. https://doi.org/10.1109/I-
SMAC52330.2021.9640923
Song, H., Zhu, E., & Ma, L. (2020). Design of
embedded real-time target tracking system
based on KNN algorithm. 443–446.
https://doi.org/10.1109/ITAIC49862.2020.9
339101
P-ISSN: 2656-1743 | E-ISSN: 2656-1735
DOI: https://doi.org/10.34288/jri.v5i2.467
JURNAL RISET INFORMATIKA
Vol. 5, No. 1 March 2022
Accredited rank 3 (SINTA 3), excerpts from the decision of the Minister of RISTEK-BRIN No. 200/M/KPT/2020
176
Sun, Y., Chen, M., & Lin, G. (2020). Design of
Intelligent Paper Counting System Based on
KNN Algorithm. Proceedings of the 32nd
Chinese Control and Decision Conference, CCDC
2020, 2910–2914.
https://doi.org/10.1109/CCDC49329.2020.9
164216
Sushmitha, K., & Jagadeesh, P. (2022). Feature
Extraction and Classification of Bone Tumor
using CNN Classifier with KNN Classifier. 2022
International Conference on Business Analytics
for Technology and Security, ICBATS 2022.
https://doi.org/10.1109/ICBATS54253.2022
.9759035
Yao, L., & Cao, T. (2022). Research on Talent
Classification Based on Improved KNN
Algorithm. Proceedings - 2022 International
Symposium on Control Engineering and
Robotics, ISCER 2022, 252–256.
https://doi.org/10.1109/ISCER55570.2022.
00050
Yunneng, Q. (2020). A new stock price prediction
model based on improved KNN. Proceedings -
2020 7th International Conference on
Information Science and Control Engineering,
ICISCE 2020, 77–80.
https://doi.org/10.1109/ICISCE50968.2020.
00026
Zhai, H. (2022). Improving KNN Algorithm
Efficiency Based on PCA and KD-tree.
Proceedings - 2022 International Conference
on Machine Learning and Knowledge
Engineering, MLKE 2022, 83–87.
https://doi.org/10.1109/MLKE55170.2022.
00021