1. Introduction
Up to now, internet technology has exploded and come into widespread use, and the number of customers shopping online is always increasing day by day (1). Companies and businesses always strive every day to satisfy consumer needs. Besides illustrators, introductory content, and product descriptions, there is also a great appearance of clothes to attract customers’ attention (2, 3). Indeed, some authors have set up a dataset of more than one hundred thousand images labeled with brands, types, colors, materials, and prices so that customers can choose the right clothes when shopping online (4–7). Additionally, many businesses have seen the strong influence of the Internet potential on consumers, which leads them to create a lot of online shopping websites such as research (8–12). On such websites, customers will choose sizes and try on clothes through a virtual simulation program which helps them fit clothes in a short time. For customers who do not have experience in purchasing clothes online, there is also a supporting website through the advice of fashion experts such as studies (13, 14). For the product to meet the needs of customers, a research group has studied to find the factors that affect the buyers’ decisions (15–17). Similarly, the method of comparing offline and online shopping behavior of buyers has been researched by some authors, and they have found out the motivating factors affecting customers’ purchases online. These factors are the customer’s favorite fashion style, and the combination of materials or colors as research results (18, 19). To create an additional direction for choosing clothes online, the content of this paper presents the use of the fuzzy logic simulation to choose Torso dress styles that match customer preferences in terms of form, color, and fabric.
2. Materials and methods
Basic Torso patterns use for draping dresses on avatars. There are five groups based on the fabrics’ thickness, and this is set according to the database of the CLO3D software. The first group is very thin fabrics including chiffon, voile, and muslin. The second is the thin ones including silk fabric and brocade fabric. The third and fourth groups are medium-weight and thick fabrics, respectively. Cotton and linen fabrics belong to medium-weight fabrics, while denim and khaki fabric belong to the thick ones. The very thick fabrics are the final group including denim fabric and khaki fabric.
The 2D Torso patterns work is designed on the Gerber Accumark software, combined with the simulation on CLO3D software to create a dress collection. To choose a Torso dress style, the author designed a simulation program by using the fuzzy logic technique. In addition, the experimental content is conducted through customer behavior surveys, which are analyzed according to the Pareto chart. Finally, the simulation modeling results have checked the expert method’s feasibility.
3. Results and discussion
3.1. Survey analysis based on customer behavior in choosing clothes
This is the first study that was conducted on 145 female customers’ positions, who are 18 to 25°years old based on five factors: form, fabric, color, decoration, and price. The survey is analyzed by Pareto chart (Figure 1) and there are three factors mostly affecting clothes consumption on the customer service website: form, fabric, and color.
Figure 1. The Pareto chart of factors affecting customer behavior.
In terms of the dress fabric, there are 7 thickness factors: medium, thin, thick, very thin, very thick, lace, and knit that exert influence on buyers’ selection. According to the chart, customers give priority to medium, thin, thick, very thin, and very thick fabrics (Figure 2).
Figure 2. The Pareto chart of factors affecting choice of dress fabric.
There are also 4 features in the field of the dress form: fit, wide, very wide, and tight dress. As a result, female prefer to choose fit and wide dresses (Figure 3).
Figure 3. The Pareto chart of factors affecting choice of dress form.
Moreover, the dress color include: neutral, hot, cool, black and white, opposite color, analog color, and mono-color. Neutral, hot, and cool are the favorite colors among the others (Figure 4).
Figure 4. The Pareto chart of factors affecting choice of dress color.
Through the survey results, the fabric type, form, and color are three factors mostly affecting the customers’ choice. These three factors are used as three input variables of the fuzzy logic simulation. Thereby, the total number of costume samples according to the fuzzy control modeling in the research is obtained by the following formula:
where
y: samples
x1: membership functions for the dress form
x2: membership functions for the dress fabric
x3: membership functions for the dress color
From formula (1), the total number of samples in the study can be calculated:
y = 2*5*3 = 30 (samples)
3.2. Coding dresses
The rules matrix of the simulation modeling is presented in Table 1. The numeric “0” doesn’t combine between the dress form and the fabric. The numeric “1” combines between the dress form and the fabric.
Table 1. Rules matrix of sizes.
Based on the presented three input variables, combining the number of modelings from (1) will have 30 samples of Torso dresses and perform sample coding as shown in Table 2.
Table 2. Rules matrix of sizes.
3.3. Coding dresses
The Torso dress collection is obtained after simulating 30 Torso patterns according to three input variables in a fuzzy logic modeling with styles in Table 3. The figure number of each type corresponds to the encoding number of that type.
Table 3. Figures of Torso dresses.
Torso patterns are designed by the 2D method of the formula system (20). Next, these patterns are draped on the avatar by CLO3D software. Measurements of body dimensions in the size chart are size 8 so the avatars sizing is edited to size 8 as well. Torso patterns include a front body, a back body, and a sleeve (Figure 35).
Figure 35. The Torso patterns (20).
3.4. Fuzzy rules
Sugeno fuzzy control modeling is used for the study. Fuzzy rules are established from membership functions in three input variables and one output result (Figure 36). The first input variable is the form of the modeling. The second is the fabric type that can be used for dress, and the color of the fabric is the final variable.
Figure 36. The fuzzy logic modeling for choosing the Torso dress.
There are two, five, and three membership functions for the first, second, and third input variables, respectively, (Figure 37). There are also 15 samples for the fit form showing from 1 to 15, and the same quantities are used for the wide form showing from 16 to 30. Every fabric group has 6 samples as well as every color has 10 samples. The first variable is the dress form in which its membership functions are the medium-dressed range [0.5 2 3 5], and the wide-dressed range [4 7 8 10]. The second variable is the fabric thickness with five membership functions representing five levels of it. Very thin and thin fabrics are [−1 0.5 1.5 3] and [2 3.5 5], respectively. Medium fabric is [4 5.5 7]. Thick and very thick fabrics are [6 7.5 9] for the former and [8 9.5 10.5 12] for the latter. The third variable is the fabric color. The range of the hot color is [−1 0.5 1.5 4], the neutral color is [2 5 8], and the cool color is [6 8.5 9.5 12].
Figure 37. The graph of membership functions for input variables. (A) The first input, (B) The second input, (C) The third input.
These variables will be combined by the fuzzy rule “If-Then” as the Figure 38.
Figure 38. The third input. The Anfis Model Structure of the Torso Dress.
3.5. Defuzzification
Through the defuzzification modeling (Figure 39), designers or customers will know the sample that is suitable for requirements of form, color, and fabric. These statistics will be compared with the figures in Table 3 to know the style of dress. This is a fit dress, medium fabric, cool color. The results are shown in Figure 40, and the flowchart of choosing the Torso dress is presented in Figure 41.
Figure 39. The Fuzzy logic controller modeling of the Torso dress.
Figure 40. The result of choosing the Torso dress.
Figure 41. The modeling for choosing the Torso dress.
3.6. Experiment to test and evaluate the modeling
A total of 6 lecturers and technical staff working at the garment company play the role running the simulation program and modeling evaluation to choose the type of compaction according to the expert method. The results after testing give 7 criteria: Observing the dress form before sewing the pattern; Observing the fabric thickness; Knowing the color of the dress fabric; Choosing a quick dress style; Easily choosing a different dress style; Suitable for online shopping; Cronbach’s Alpha coefficient is greater than 0.7. In conclusion, the study results are reliable (Table 4). The limitation of the study was that it did not involve many kinds of fabrics with different patterns.
Table 4. Figures of Torso dresses.
4. Conclusion
The study has established a simulation model to choose a Torso dresses style by the fuzzy logic technique. The modeling requires 3 input variables which are form, thickness, and color of the fabric, as well as the output variables result from Torso-style. The fuzzy If-Then rule and Sugeno controller were used in this modeling. In addition, there were 30 Torso dress styles coded for application to fuzzy logic modeling. A total of 145 people took part in a survey conducted by the author about their priorities when choosing new clothes or new sewing according to questions related to 3 input variables. Among the takers, there were two lecturers teaching in Garment Technology, and four people were working in the technical departments of garment companies. Moreover, this modeling has also been tested and evaluated by expert methods. This research opens up the direction of choosing a style completely different from the traditional method. Customers or designers only need to enter the parameters of movement according to the width of fabric form, color, and thickness. Then the favorite color group will know the code number of the dress. This result will be compared with the collection album to know the right dress style. The choosing clothes by the fuzzy logic method is suitable for current trend in the strong development of online clothing shopping. This study has practical applications in the production and trading of costumes and is in line with the current 4.0 trend.
Conflict of interest
The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
References
1. Beckers J, Barbosa IDC, Verhetsel A. Identifying the geography of online shopping adoption in Belgium. J Retail Consum Serv. (2018) 45:33–41. doi: 10.1016/j.jretconser.08.006
2. Jaller M, Pahwa A. Evaluating the environmental impacts of online shopping: a behavioral and transportation approach. Transport Res D Transport Environ. (2020) 80:10. doi: 10.1016/j.trd.2020.102223
3. Suleman D, Zuniarti I. Consumer decisions toward fashion product shopping in Indonesia: the effects of attitude. Percept Ease Use Usefulness Trust Manag Dyn Knowl Econ. (2019) 7:143–6. doi: 10.25019/MDKE/7.2.01
4. Liu KH, Fei W, Liu TJ. A clothing recommendation dataset for online shopping. Proceedings of the IEEE International Conference on Consumer Electronics - Taiwan (ICCE-TW). Yilan (2019). p. 1–2. doi: 10.1109/ICCE-TW46550.2019.8991796
5. Ladhari R, Gonthier J, Lajante M. Generation Y and online fashion shopping: orientations and profiles. J Retail Consum Serv. (2019) 48:113–21. doi: 10.1016/j.jretconser.2019.02.003
6. Redda EH. Attitudes towards online shopping: application of the theory of planned behaviour. Acta Univ Danubius. (2019) 15:148–59.
7. Ah PC, Kai YC. Improving the apparel virtual size fitting prediction under psychographic characteristics and 3D body measurements using artificial neural network. Hum Fact Apparel Textile Eng. (2022) 2332:94–100. doi: 10.54941/ahfe1001543
8. Lokman AM, Noor L, Nagamachi M. Dominant pattern of visual design in online clothing websites. IEEE Comput Graph Applic. (2003) 23:38–48. doi: 10.1109/MCG.2003.1159612
9. Wang Y, Fu W, Wang Y, Xianglin H. Retrieval of clothing images based on color feature. Amsterdam: Atlantis Press (2015). p. 143–9. doi: 10.2991/amcce-15.2015.26
10. Nada YA, Meshref H. Analysis, design, and implementation of intelligent expert system for clothes style selection. Int J Comput Applic. (2014) 105:15–20.
11. Petrosova IA, Andreeva EG, Guseva MA. The system of selection and sale of ready-to-wear clothes in a virtual environment. Proceedings of the International Science and Technology Conference “EastConf”. Vladivostok (2019). p. 1–5. doi: 10.1109/EastConf.2019.8725390
12. Jingsi Z. Research on the application of 3D virtual simulation technology in fashion design from the perspective of meta universe. Sci Soc Res. 4:19–23.
13. Sekozawa T, Mitsuhashi H. One-to-one recommendation system in apparel online shopping. Electr Commun Japan. (2016) 94:51–60. doi: 10.1002/ecj.10261
14. Sumarliah E, Usmanova K, Mousa K, Indriya I. E-commerce in the fashion business: the roles of the COVID-19 situational factors, hedonic and utilitarian motives on consumers’ intention to purchase online. Int J Fashion Design Technol Educ. (2021) 15:167–77. doi: 10.1080/17543266.2021.1958926
15. Katawetawaraks C, Wang CL. Online shopper behavior: influences of online shopping decision. Asian J Bus Res. (2011) 1:66–74. doi: 10.14707/ajbr.110012
16. Sithole PPN. Factors Affecting Purchase Decisions of Online Shopping for Clothing”, Ph.D. Thesis. (2013).
17. Merta PGBMI, Trinanasari N, Bagia WI. Influential factors of online shopping decision. Int J Soc Sci Bus. (2022) 6:246–9. doi: 10.23887/ijssb.v6i2.44077
18. Neacsu NA, Codruta BS. “Adina Baltescu, DANA, Boşcor, The influence of design and aesthetics elements in choosing clothing. Industr Textila. (2017) 68:375–9.
19. Cowart K, Goldsmith R. The Influence of Consumer Decision-Making Styles on Online Apparel Consumption by College Students. New York, NY: Wiley Online Library (2007). doi: 10.1111/j.1470-6431.2007.00615.x