Vietnamese females between the ages of 25 and 35. Determine the sample size according to the following formula (1):

n=t2×σ2m2=1.962× 2.9120.52=130⁢(people)            (1)  (15)

In Equation (1), n is the sample (people); t is the Student coefficient (t = 1.96); m is error (m = 0.05); and σ is standard deviation (σ = 2.91 cm). Thus, the sample size selected for this study has an accuracy of 95% and requires 130 people.

The research uses Gerber Accumark two-dimensional (2D) and V-Stitcher 3D software. Besides that, lace fabric is used to sew bras.

The method of data collection: directly measuring 24 anthropometric dimensions, including one height, three body length dimensions, 14 body circumference dimensions, and six bust-related dimensions. Body shape analysis techniques: Using SPSS 21.0 software to perform principal component analysis of a set of dimensions to reduce large amounts of data into a small number of subgroups with common features. It uses analytical subgroups of subjects by K-means and discriminant analysis to find the differences between groups of research subjects (16, 17). Finally, it is used to test analysis of variance (ANOVA) for the mean of independent sample groups. Based on this test result and FFIT’s criteria to classify body shape, the bras are created using a 2D pattern from the basic block (18) and the bra design formula system (19). The results of the study are tested for fit on a 3D simulated method through the tension map and pressure map. They are checked for reliability according to Cronbach’s alpha coefficient.

Through a survey of 24 body sizes of 130 women in the study age group to analyze the body shapes by the K-mean cluster method. The results show the mean and standard deviation (SD) of the four body groups as follows (Table 1). The body’s dimensions are measured as in Figure 1.

The results of the classification of 130 samples into three kinds of groups are as follows: there are three groups, four groups, and five groups. Besides that, K-mean cluster analysis allows the analyst to proactively impose the number of groups to be classified. Based on the results of K-mean cluster analysis, initially choosing the solution divided into three groups, four groups are the most suitable because the distribution chart of these groups has no overlap between the groups (Figure 2). Combining the ANOVA test, describing the data dispersion, the solution divided into four groups was selected. In the solution for four groups, the difference between variables is the highest compared to the three group solution. The solution of dividing into four groups with 23/27 variables with a Sig < 0.05 value shows that there is a quantitative difference between the groups being compared. Based on classification criteria for FFIT body shape, all groups are simulated on virtual avatars. Groups 1 (Figure 3) and Group 3 (Figure 4) have oval shaped bodies. Compared with Group 3, Group 1 had wider shoulders, bust, and waist circumference. In Group 3, the waist was not clearly distinguished from the buttock circumference. Group 2 (Figure 5) has an hourglass shape with a slight difference in hip and bust measurements of about 6.65 cm. The waist measurement (64.27 cm) is clearly distinguished from that of the hip (86.88 cm) and the bust (80.23). Group 4 (Figure 6) has a rectangular shape. The measurements of the hip (79.5 cm) and the bust (77.5 cm) are not significantly different. The waist (63 cm) is not clearly distinguished compared with the buttocks and the bust.

These different variables are the characteristic dimensions of the upper body: bust, waist, shoulder width, cross-chest, and bust height. The dimension components that need attention when designing samples, which have a great impact on the body analysis process for Vietnamese women aged 25–35, are shoulder width, across front/back, bust, waist, and hip. Based on the chest shape profile, it shows that the busts of four groups have different average bust sizes, but there is not much difference in bust shape. In Group 1, the average bust measurement was 88.6 cm, Group 3 was 84 cm, and the difference in bust between the two groups was 4.6 cm. Group 2 has an average bust measurement of 80 cm, while Group 4 is 77 cm. The difference in bust between the two groups is 3 cm. Group 2 (purple color) and Group 4 (blue color) have relatively similar bust sizes, being smaller than Group 1 (black color) and Group 3 (red color) (Figure 7).

Based on Table 1, Group 3 is the group with the average measurement among the four groups and also the group with the largest number of people (39%), so Group 3 will be selected as the base group for flat pattern design (Table 2).

Fit clothes are all about the feeling of satisfaction when wearing them. Figure 8 shows three designs worn on models with base group body measurements. The survey of satisfaction about wearing feelings was conducted on 10 people with bust measurements belonging to the base group, with six criteria: comfort, beautiful design, like to use, and easy to wear, round fit chest, and easy movement. They all agree with these six criteria. Then, calculate Cronbach’s alpha with SPSS to evaluate the reliability of the resulting scale. The results of five samples show all greater than 0.7 (Table 3). This shows that the aforementioned survey results are completely reliable; there is a correlation and consistency in the answers of the respondents.

In addition, the authors use 3D V-Stitcher software to test the fit of the bras on the virtual avatar (Figure 9). The grid areas are green and yellow, showing that the designs fit perfectly and bring comfort and ease to daily use.