Available via license: CC BY 3.0
Content may be subject to copyright.
IOP Conference Series: Materials Science and Engineering
PAPER • OPEN ACCESS
Supplementary quality control features for the
production department in Odoo ERP
To cite this article: A S Ahmadiyah et al 2021 IOP Conf. Ser.: Mater. Sci. Eng. 1072 012055
View the article online for updates and enhancements.
You may also like
Lifetime of topological quantum memories
in thermal environment
Abbas Al-Shimary, James R Wootton and
Jiannis K Pachos
-
Bounding quantum gate error rate based
on reported average fidelity
Yuval R Sanders, Joel J Wallman and
Barry C Sanders
-
An automated, quantitative, and case-
specific evaluation of deformable image
registration in computed tomography
images
R G J Kierkels, L A den Otter, E W
Korevaar et al.
-
This content was downloaded from IP address 168.151.109.206 on 04/02/2023 at 17:31
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd
IConISE-ACISE 2020
IOP Conf. Series: Materials Science and Engineering 1072 (2021) 012055
IOP Publishing
doi:10.1088/1757-899X/1072/1/012055
1
Supplementary quality control features for the production
department in Odoo ERP
A S Ahmadiyah1,2, Y Y Ratna1,3, N N Yotifa1,4, and I Dinillah1,5
1Informatics Department, Sepuluh Nopember Institute of Technology, Surabaya,
Indonesia
E-mail: 2adhatus@if.its.ac.id, 3yukiratna.17051@mhs.its.ac.id,
4nitama.17051@mhs.its.ac.id, 5izzahdinillah.17051@mhs.its.ac.id
Abstract. Odoo is one of the top Enterprise Resource Planning (ERP) applications. In particular,
it is supported by a quality control module with the ability to control, trigger an alert, and check
its purpose. This research strives to supplement the existing Odoo quality control module with
parameters from past inspection data obtained before quality control migrates to Odoo. To
support the study, we used a machine learning method to discover the intrinsic pattern from the
dataset of quality control in the production process of baby biscuit products. The experiment
shows that the additional quality control feature can provide the product measurement and
tolerance threshold fed into Odoo quality control module. The additional feature is helpful for
decision making and error minimization in setting quality control parameters and tolerance
threshold. Furthermore, a high accuracy rate of 95.71% is obtained from the employed Decision
Tree algorithm.
1. Introduction
Quality control (QC) is vital to ensure that the final product or service meets the requirements and a set
of quality standards. It supports companies to maintain and protect the quality of products or services,
which leads to achieving the target market opportunity through customer satisfaction [1]. Specifically,
in the manufacturing industry, quality control appears as quality control of raw materials, quality control
in the production process, and quality control of the final product.
In a large-scale business, quality control is recorded into an integrated management system such as
an Enterprise Resource Planning (ERP). Odoo is one of the popular open-source ERP tools. Odoo is
equipped with quality inspection in which one can create an inspection plan based on business processes
and company needs. The role of ERP in quality inspection is to speed up and simplify the inspection
process.
Odoo quality control consists of three functions, i.e., control, check, and alert functions. In defining
the quality control point using take measure type, one needs to input product measurement along with
the threshold value during the production process. This feature can only be used for the currently
inspected data. However, this feature does not accommodate past data before the company migrates
quality control to Odoo, which potentially holds essential information about measurement and threshold.
This paper aims to tackle this problem by suggesting additional quality control features by extracting
inherent information from the past data. We utilized a machine learning method to enable the learning
process. Machine learning is a powerful tool to conduct quality inspection [2][3][4][5][6][7]. Our
proposed solution can automatically find the product measurement and threshold values fed to the Odoo
quality control module by using the baby biscuit production process in our experiment.
2. Methodology
Our concern is to incorporate essential information from past inspection data generated before the QC
migration to Odoo as a recommendation fed to define quality control points. As seen in Figure 1, our
proposed process is illustrated in the bottom left side. It starts with the preparation of the past inspection
dataset. Then, a Decision Tree algorithm is utilized to discover intrinsic patterns. Finally, the
IConISE-ACISE 2020
IOP Conf. Series: Materials Science and Engineering 1072 (2021) 012055
IOP Publishing
doi:10.1088/1757-899X/1072/1/012055
2
recommended product measurement and threshold from the past is obtained. The term past refers to
quality control conducted without the Odoo quality control module.
Figure 1. The position of our proposed process against the existing Odoo quality control module.
As seen in Table 1, the past inspection dataset from baby biscuit is prepared. From 233 collected
records, the quality control dataset has several dimensional measurements or attributes. They are shape,
diameter in centimeters, weight in grams, and color in RGB, separated into three-color channels ranging
from zero to 255. The products are divided into five codes; there are M01, M02, M03, M04, and M05;
each represents the flavor of the baby biscuits.
Table 1. Samples of baby biscuit quality control in production process.
Product
Code
Circular
Shape
Diameter
(cm)
Weight
(gm)
Color Channel
Red
Green
Blue
M03
Yes
5.9
9
191
173
109
M02
Yes
5.8
8.9
189
179
98
M04
Yes
6
10.83
192
161
101
M01
Yes
4.9
10.83
193
173
111
M05
No
5.6
11
213
182
101
M05
Yes
5.7
12.1
190
162
101
M04
Circle
6
8
189
175
110
M04
Circle
5.8
12.5
201
170
104
Next, the discovering intrinsic patterns step is performed using the Iterative Dichotomiser 3 (ID3)
algorithm [8], one type of the Decision Tree, to train each data in the dataset. The ID3 algorithm
generates a tree structure. At first, The ID3 selects the root node, which has the highest information gain
value. It is followed by setting the next attribute with the highest information gain value among the rest
of the attributes as the branch node. This process is repeated until all attributes are counted.
IConISE-ACISE 2020
IOP Conf. Series: Materials Science and Engineering 1072 (2021) 012055
IOP Publishing
doi:10.1088/1757-899X/1072/1/012055
3
From this step, we have a model to classify each data into pass or defect type. This model's
performance is then tested to ensure the accuracy of the model, leading to trusted results fed into the
Odoo quality control module. Lastly, we calculate product measurement tolerance from the model built,
which is also needed by the Odoo quality control module.
3. Results and Discussions
From 233 records used in the training phase, the classifier can predict 95.71% of the quality control
status correctly, leaving the 4.29% misclassified. Meanwhile, Table 2 shows the result of the testing
dataset using the model built using the ID3 algorithm. The column labeled Expected Output is the result
of the past inspection. In contrast, the column labeled Real Output represents the result obtained using
ID3. In this case, The ID3 algorithm is proved to perform well in revealing the intricate pattern of the
dataset.
Table 2. Quality control test results.
Test
#
Product
Code
Circular
Shape
Diameters
(cm)
Weight
(gm)
Color Channel
Expected
Output
Real
Output
Red
Green
Blue
1
M03
Yes
5.6
10
203
177
101
Pass
Pass
2
M02
Yes
5.8
9
210
172
101
Pass
Pass
3
M04
Yes
5.7
11
205
181
109
Pass
Pass
4
M01
Yes
4.9
10
201
170
109
Defect
Defect
5
M05
No
5.8
9
202
180
109
Defect
Defect
6
M05
Yes
5.6
10
216
170
105
Defect
Defect
7
M04
Yes
5.7
8
201
171
102
Defect
Defect
8
M01
No
5.8
9
191
179
109
Defect
Defect
In Table 3, we organized product measurement and tolerance threshold similar to the Odoo quality
control module, particularly defining the quality control point feature. The product measurement name,
type, and unit of measurement (UoM) are obtained directly from the dataset attributes. In contrast, the
minimum, maximum, and tolerance value of the quantitative type are obtained from machine learning,
in our case, the ID3 algorithm. Holding that information, the quality control team has the option to
directly use the recommendation from our supplementary quality control feature or make adjustments.
This can lead to minimizing the effort of setting quality control points and human error.
Table 3. Product measurement and tolerance threshold.
Name
Type
Minimum
tolerance
Minimum
Maximum
Maximum
tolerance
UoM
Qualitativ
e value
Circular shape
qualitative
0
0
0
0
-
2 records
(Yes, No)
Diameters
quantitative
0.2
5.8
5.8
0.2
centimeters
-
Weight
quantitative
2
10.83
10.83
2
grams
-
Red channel
quantitative
0
185
220
0
-
-
Green channel
quantitative
0
140
165
0
-
-
Blue channel
quantitative
0
0
80
0
-
-
To be used by other companies, this solution can be adopted directly. However, the high accuracy of
machine learning may vary due to the inspection of the dataset's characteristics. The accuracy of the
IConISE-ACISE 2020
IOP Conf. Series: Materials Science and Engineering 1072 (2021) 012055
IOP Publishing
doi:10.1088/1757-899X/1072/1/012055
4
model built under different inspection dataset needs to be checked before continuing to be filled into the
Odoo quality control module.
4. Conclusion
When the quality control is established long before migrating to Odoo, the past quality control data can
be extracted automatically using a machine learning method, specifically the ID3 algorithm. Usually,
the accuracy rate is the one sought. Furthermore, the splitting criteria from the ID3 algorithm are
beneficial in providing the tolerance threshold.
References
[1] Tata J, Motwani J and Prasad S. Benchmarking Quality Management Practices: US Versus Costa
Rica. Multinatl. Bus. Rev. 2000;8(2):37-42. Retrieved from:
https://www.researchgate.net/publication/292017239_Benchmarking_quality_management_
practices_US_versus_Costa_Rica
[2] Anand S, Priya L. A Guide for Machine Vision in Quality Control. Chapman and Hall/CRC; 2019
Dec 23; doi.org/10.1201/9781003002826
[3] Mohammadi P, Wang ZJ. Machine learning for quality prediction in abrasion-resistant material
manufacturing process. 2016 IEEE Can Conf Electr. Comput. Eng. (CCECE). 2016 May;
doi.org/10.1109/ccece.2016.7726783
[4] Ordukaya E, Karlik B. Quality Control of Olive Oils Using Machine Learning and Electronic
Nose. J. Food Qual. 2017;2017:1–7. doi.org/10.1155/2017/9272404
[5] Alonzo LMB, Chioson FB, Co HS, Bugtai NT, Baldovino RG. A Machine Learning Approach
for Coconut Sugar Quality Assessment and Prediction. 2018 IEEE 10th International
Conference on Humanoid. Nanotechnology. Inf. Technol, Commun Control. Environ. Manag
(HNICEM). 2018 Nov; doi.org/10.1109/hnicem.2018.8666315
[6] Li F, Wu J, Dong F, Lin J, Sun G, Chen H, et al. Ensemble Machine Learning Systems for the
Estimation of Steel Quality Control. 2018 IEEE Int Conf on Big Data (Big Data). 2018 Dec;
doi.org/10.1109/bigdata.2018.8622583
[7] Bordekar L, Velingkar H, Fernandes E, Bandekar HH, Harmalkar AG, Antonio Pinto BJ. Cashew
Nut Grade Identification and Quality Testing Using Machine Learning. 2018 Second Int. Conf.
Inventive Commun. Comput. Technol (ICICCT). 2018 Apr;
doi.org/10.1109/icicct.2018.8473058
[8] Quinlan, J.R. Induction of decision trees. Mach Learn 1. 1986:81–106.
doi.org/10.1007/BF00116251. Retrieved from:
https://link.springer.com/article/10.1007/BF00116251
